wav
14.4 KBASR Cue: “one” Clean (WAV)
Short synthetic cue tone for “one”: clean reference for wake-word / digit ASR smoke tests.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
Audio
Audio test files usually arrive as someone's music clip. These are engineered. Silence sets document leading and trailing trim points; ASR suites pair synthetic digit utterances with transcripts and noise twins; loudness ladders, peak-clip pairs, stereo/mono, sample-rate conversion, and DTMF sequences exercise meters and harnesses. Every file is short to stay within budget and ships with documented sample rate, bit depth, and duration.
Filter audio on Browse · 396 files · 18 subcategories
Frequently asked questions
Are the ASR clips real speech recordings?
No, Wave B ASR fixtures are synthetic tone-digit utterances with JSON transcripts. They exercise ASR harnesses and loaders without shipping copyrighted speech.
What are the silence trim fixtures for?
Documented leading/trailing silence (1–5 s) on pure tones so auto-trim tools can be checked against exact timestamps. Wave E also adds mid-gap silence cases for split/trim tools.
Which formats are included?
WAV references plus compressed codecs (MP3, FLAC, OGG, and more). Specs list sample rate, bit depth, and duration.
How do I test loudness meters or peak clip detection?
Filter Browse by purpose loudness-testing for the amplitude ladder and clean↔hard-clipped peak pairs. Stereo/mono and sample-rate twins cover channel and resampler QA.
396 of 396 files
Asr
wav
14.4 KBASR Cue: “one” Noisy (WAV)
Noisy twin of the “one” cue tone (~5 dB SNR).
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
14.4 KBASR Cue: “zero” Clean (WAV)
Short synthetic cue tone for “zero”: clean reference for wake-word / digit ASR smoke tests.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
14.4 KBASR Cue: “zero” Noisy (WAV)
Noisy twin of the “zero” cue tone (~5 dB SNR).
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
27.9 KBASR Extra: 0000 Clean (WAV)
Extra synthetic digit utterance 0000 (zero zero zero zero): clean twin.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
27.9 KBASR Extra: 0000 Noisy (WAV)
Noisy twin of digit sequence 0000.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
165 BASR Extra: 0000 Transcript (JSON)
Ground-truth transcript for extra ASR utterance 0000.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
27.9 KBASR Extra: 1111 Clean (WAV)
Extra synthetic digit utterance 1111 (one one one one): clean twin.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
27.9 KBASR Extra: 1111 Noisy (WAV)
Noisy twin of digit sequence 1111.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
161 BASR Extra: 1111 Transcript (JSON)
Ground-truth transcript for extra ASR utterance 1111.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
27.9 KBASR Extra: 2222 Clean (WAV)
Extra synthetic digit utterance 2222 (two two two two): clean twin.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
27.9 KBASR Extra: 2222 Noisy (WAV)
Noisy twin of digit sequence 2222.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
161 BASR Extra: 2222 Transcript (JSON)
Ground-truth transcript for extra ASR utterance 2222.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
27.9 KBASR Extra: 3333 Clean (WAV)
Extra synthetic digit utterance 3333 (three three three three): clean twin.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
27.9 KBASR Extra: 3333 Noisy (WAV)
Noisy twin of digit sequence 3333.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
169 BASR Extra: 3333 Transcript (JSON)
Ground-truth transcript for extra ASR utterance 3333.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
27.9 KBASR Extra: 4444 Clean (WAV)
Extra synthetic digit utterance 4444 (four four four four): clean twin.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
27.9 KBASR Extra: 4444 Noisy (WAV)
Noisy twin of digit sequence 4444.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
165 BASR Extra: 4444 Transcript (JSON)
Ground-truth transcript for extra ASR utterance 4444.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
27.9 KBASR Extra: 5555 Clean (WAV)
Extra synthetic digit utterance 5555 (five five five five): clean twin.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
27.9 KBASR Extra: 5555 Noisy (WAV)
Noisy twin of digit sequence 5555.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
165 BASR Extra: 5555 Transcript (JSON)
Ground-truth transcript for extra ASR utterance 5555.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
27.9 KBASR Extra: 6666 Clean (WAV)
Extra synthetic digit utterance 6666 (six six six six): clean twin.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
27.9 KBASR Extra: 6666 Noisy (WAV)
Noisy twin of digit sequence 6666.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
161 BASR Extra: 6666 Transcript (JSON)
Ground-truth transcript for extra ASR utterance 6666.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
27.9 KBASR Extra: 7777 Clean (WAV)
Extra synthetic digit utterance 7777 (seven seven seven seven): clean twin.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
27.9 KBASR Extra: 7777 Noisy (WAV)
Noisy twin of digit sequence 7777.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
169 BASR Extra: 7777 Transcript (JSON)
Ground-truth transcript for extra ASR utterance 7777.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
27.9 KBASR Extra: 8888 Clean (WAV)
Extra synthetic digit utterance 8888 (eight eight eight eight): clean twin.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
27.9 KBASR Extra: 8888 Noisy (WAV)
Noisy twin of digit sequence 8888.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
169 BASR Extra: 8888 Transcript (JSON)
Ground-truth transcript for extra ASR utterance 8888.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
27.9 KBASR Extra: 9999 Clean (WAV)
Extra synthetic digit utterance 9999 (nine nine nine nine): clean twin.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
27.9 KBASR Extra: 9999 Noisy (WAV)
Noisy twin of digit sequence 9999.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
165 BASR Extra: 9999 Transcript (JSON)
Ground-truth transcript for extra ASR utterance 9999.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
37.5 KBASR: 0123 Clean (WAV)
Synthetic clean tone sequence encoding digits 0123 (zero one two three). Pair with the noisy twin and transcript JSON for ASR evaluation.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
37.5 KBASR: 0123 Noisy (WAV)
Noisy twin of the digit sequence 0123 at ~7 dB SNR. Score ASR against the shared transcript JSON.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
245 BASR: 0123 Transcript (JSON)
Ground-truth transcript for the 0123 ASR utterance pair, expected text: “zero one two three”.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
37.5 KBASR: 1357 Clean (WAV)
Synthetic clean tone sequence encoding digits 1357 (one three five seven). Pair with the noisy twin and transcript JSON for ASR evaluation.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
37.5 KBASR: 1357 Noisy (WAV)
Noisy twin of the digit sequence 1357 at ~7 dB SNR. Score ASR against the shared transcript JSON.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
244 BASR: 1357 Transcript (JSON)
Ground-truth transcript for the 1357 ASR utterance pair, expected text: “one three five seven”.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
45.7 KBASR: 24680 Clean (WAV)
Synthetic clean tone sequence encoding digits 24680 (two four six eight zero). Pair with the noisy twin and transcript JSON for ASR evaluation.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
45.7 KBASR: 24680 Noisy (WAV)
Noisy twin of the digit sequence 24680 at ~7 dB SNR. Score ASR against the shared transcript JSON.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
249 BASR: 24680 Transcript (JSON)
Ground-truth transcript for the 24680 ASR utterance pair, expected text: “two four six eight zero”.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
37.5 KBASR: 4567 Clean (WAV)
Synthetic clean tone sequence encoding digits 4567 (four five six seven). Pair with the noisy twin and transcript JSON for ASR evaluation.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
37.5 KBASR: 4567 Noisy (WAV)
Noisy twin of the digit sequence 4567 at ~7 dB SNR. Score ASR against the shared transcript JSON.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
246 BASR: 4567 Transcript (JSON)
Ground-truth transcript for the 4567 ASR utterance pair, expected text: “four five six seven”.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
21.3 KBASR: 89 Clean (WAV)
Synthetic clean tone sequence encoding digits 89 (eight nine). Pair with the noisy twin and transcript JSON for ASR evaluation.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
21.3 KBASR: 89 Noisy (WAV)
Noisy twin of the digit sequence 89 at ~7 dB SNR. Score ASR against the shared transcript JSON.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
233 BASR: 89 Transcript (JSON)
Ground-truth transcript for the 89 ASR utterance pair, expected text: “eight nine”.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
wav
53.8 KBASR: 987654 Clean (WAV)
Synthetic clean tone sequence encoding digits 987654 (nine eight seven six five four). Pair with the noisy twin and transcript JSON for ASR evaluation.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
53.8 KBASR: 987654 Noisy (WAV)
Noisy twin of the digit sequence 987654 at ~7 dB SNR. Score ASR against the shared transcript JSON.
- File
- WAV · Asr · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
258 BASR: 987654 Transcript (JSON)
Ground-truth transcript for the 987654 ASR utterance pair, expected text: “nine eight seven six five four”.
- File
- JSON · Asr
- Use case
- ASR testingJSON parsing· Conversion set
Bit Depth Ladder
wav
172.3 KB440 Hz Tone @ 16-bit (WAV)
A 440 Hz sine tone stored as 16-bit PCM: part of a bit-depth ladder (8/16/24-bit) of the identical tone, for testing bit-depth conversion, dithering, and quantisation-noise handling.
- File
- WAV · Bit Depth Ladder · 2 s
- Use case
- Audio analysisConversion testing· Conversion set
wav
258.4 KB440 Hz Tone @ 24-bit (WAV)
A 440 Hz sine tone stored as 24-bit PCM: part of a bit-depth ladder (8/16/24-bit) of the identical tone, for testing bit-depth conversion, dithering, and quantisation-noise handling.
- File
- WAV · Bit Depth Ladder · 2 s
- Use case
- Audio analysisConversion testing· Conversion set
wav
86.2 KB440 Hz Tone @ 8-bit (WAV)
A 440 Hz sine tone stored as 8-bit PCM: part of a bit-depth ladder (8/16/24-bit) of the identical tone, for testing bit-depth conversion, dithering, and quantisation-noise handling.
- File
- WAV · Bit Depth Ladder · 2 s
- Use case
- Audio analysisConversion testing· Conversion set
Channels
wav
3 MB5.1 Surround Channel Identification (WAV, 6ch)
A six-channel 5.1 surround WAV that plays a 1-second tone in each channel in turn, in the standard WAV order (FL, FR, C, LFE, SL, SR). A fixture for testing multichannel decoding, downmixing, and channel-order handling.
- File
- WAV · Channels · 44100 Hz
- Use case
- Audio analysis
wav
516.8 KBStereo Channel Map: L 440 Hz / R 880 Hz (WAV)
A stereo WAV with a distinct tone in each channel (440 Hz on the left, 880 Hz on the right), so you can verify channel mapping, panning, and L/R routing unambiguously.
- File
- WAV · Channels · 3 s
- Use case
- Audio analysis
Codec Set
aac
71.9 KBAAC: ADTS
The clip as raw AAC in an ADTS stream: the codec behind most streaming and mobile audio. For testing AAC decoders and remux into MP4.
- File
- AAC · Codec Set · AAC-LC
- Use case
- Conversion testingAudio analysis· Conversion set
ac3
71 KBAC-3: Dolby Digital
The clip as AC-3 (Dolby Digital): the multichannel codec used in DVD/broadcast. Rendered here in stereo; for testing AC-3 decoding and conversion.
- File
- AC3 · Codec Set · AC-3 (Dolby Digital)
- Use case
- Conversion testingAudio analysis· Conversion set
m4a
184.4 KBALAC: Apple Lossless
The clip as Apple Lossless (ALAC) in an M4A container: lossless, unlike the AAC M4A twin. For testing ALAC decoding and lossless conversion.
- File
- M4A · Codec Set · ALAC
- Use case
- Conversion testingAudio analysis· Conversion set
amr
4.7 KBAMR: Narrowband Speech
The clip resampled to 8 kHz mono and encoded as AMR narrowband: the telephony/voice-note codec. For testing AMR decoding and speech-codec conversion.
- File
- AMR · Codec Set · AMR-NB
- Use case
- Conversion testingAudio analysis· Conversion set
flac
102.2 KBFLAC: Lossless
The clip as FLAC: free lossless audio compression. Byte-for-byte recoverable to the source PCM; for testing lossless decoders and conversion.
- File
- FLAC · Codec Set · FLAC
- Use case
- Conversion testingAudio analysis· Conversion set
m4a
72.3 KBM4A: AAC (MP4)
The clip as AAC in an MP4/M4A container (faststart): Apple's default audio container. For testing M4A parsing and MP4 audio conversion.
- File
- M4A · Codec Set · AAC-LC
- Use case
- Conversion testingAudio analysis· Conversion set
m4r
72.3 KBM4R: iPhone Ringtone
The clip as an M4R iPhone ringtone: AAC in an MP4 container with the ringtone extension. For testing that a converter maps M4R↔M4A correctly.
- File
- M4R · Codec Set · AAC-LC
- Use case
- Conversion testingAudio analysis· Conversion set
mp3
71.7 KBMP3: CBR 192 kbps
The source clip as constant-bitrate MP3 (LAME, 192 kbps): the most universally supported lossy audio format. For testing MP3 decoders, players, and conversion.
- File
- MP3 · Codec Set · MP3 (LAME)
- Use case
- Conversion testingAudio analysis· Conversion set
mp3
35.1 KBMP3: VBR (V2)
The same clip as variable-bitrate MP3 (LAME V2), for testing VBR handling, seeking, and duration estimation against the CBR twin.
- File
- MP3 · Codec Set · MP3 (LAME)
- Use case
- Conversion testingAudio analysis· Conversion set
ogg
16 KBOGG: Vorbis
The clip as Ogg Vorbis: a royalty-free lossy codec. Browser-playable; for testing Vorbis decoding and Ogg conversion.
- File
- OGG · Codec Set · Vorbis
- Use case
- Conversion testingAudio analysis· Conversion set
opus
51.9 KBOpus
The clip as Opus: the modern low-latency codec used by WebRTC and streaming. Browser-playable; for testing Opus decoding and conversion.
- File
- OPUS · Codec Set · Opus
- Use case
- Conversion testingAudio analysis· Conversion set
wav
516.8 KBWAV: Codec-Set Source
The lossless PCM WAV source for the audio conversion set: the same 3-second tone every other codec in this group is encoded from. Use it as the reference when diffing encoders.
- File
- WAV · Codec Set · PCM
- Use case
- Conversion testingAudio analysis· Conversion set
wma
103.7 KBWMA: Windows Media Audio
The clip as Windows Media Audio (WMA v2) in an ASF container: Microsoft's lossy codec. For testing WMA decoding and conversion to open formats.
- File
- WMA · Codec Set · WMA v2
- Use case
- Conversion testingAudio analysis· Conversion set
Container Formats
aiff
258.5 KBAIFF: 1 kHz Sine Tone
A pure 1 kHz sine tone stored as AIFF: Apple's big-endian 16-bit PCM container, 3 seconds, 44.1 kHz mono. A clean reference for testing AIFF decoders and WAV↔AIFF conversion.
- File
- AIFF · Container Formats · 3 s
- Use case
- Audio analysisConversion testing
aiff
258.5 KBAIFF: 440 Hz Sine Tone
A pure 440 Hz sine tone stored as AIFF: Apple's big-endian 16-bit PCM container, 3 seconds, 44.1 kHz mono. A clean reference for testing AIFF decoders and WAV↔AIFF conversion.
- File
- AIFF · Container Formats · 3 s
- Use case
- Audio analysisConversion testing
au
129.2 KBAU: 1 kHz Sine Tone
A pure 1 kHz sine tone stored as a Sun/NeXT AU file, big-endian 16-bit PCM, 3 seconds, 44.1 kHz mono. A compact reference for AU decoding and format conversion.
- File
- AU · Container Formats · 3 s
- Use case
- Audio analysisConversion testing
au
129.2 KBAU: 440 Hz Sine Tone
A pure 440 Hz sine tone stored as a Sun/NeXT AU file, big-endian 16-bit PCM, 3 seconds, 44.1 kHz mono. A compact reference for AU decoding and format conversion.
- File
- AU · Container Formats · 3 s
- Use case
- Audio analysisConversion testing
Dtmf
wav
254.1 KBDTMF Dialing Tones (WAV)
A DTMF (touch-tone) dialing sequence dialing a reserved 555-01xx number: each digit is the standard dual-tone pair. A fixture for testing DTMF decoders, Goertzel detectors, and tone analysis.
- File
- WAV · Dtmf · 44100 Hz
- Use case
- Audio analysis
wav
21.9 KBDTMF Sequence: *9#
Synthetic DTMF tone sequence for digits “*9#”. Pair with the transcript JSON.
- File
- WAV · Dtmf · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
21.9 KBDTMF Sequence: 042
Synthetic DTMF tone sequence for digits “042”. Pair with the transcript JSON.
- File
- WAV · Dtmf · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
28.2 KBDTMF Sequence: 1234
Synthetic DTMF tone sequence for digits “1234”. Pair with the transcript JSON.
- File
- WAV · Dtmf · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
34.4 KBDTMF Sequence: 13579
Synthetic DTMF tone sequence for digits “13579”. Pair with the transcript JSON.
- File
- WAV · Dtmf · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
wav
40.7 KBDTMF Sequence: 567890
Synthetic DTMF tone sequence for digits “567890”. Pair with the transcript JSON.
- File
- WAV · Dtmf · 16000
- Use case
- ASR testingAudio analysis· Paired fixture
json
224 BDTMF Transcript: *9# (JSON)
Ground-truth digit transcript for the DTMF sequence *9#.
- File
- JSON · Dtmf
- Use case
- ASR testing· Paired fixture
json
224 BDTMF Transcript: 042 (JSON)
Ground-truth digit transcript for the DTMF sequence 042.
- File
- JSON · Dtmf
- Use case
- ASR testing· Paired fixture
json
227 BDTMF Transcript: 1234 (JSON)
Ground-truth digit transcript for the DTMF sequence 1234.
- File
- JSON · Dtmf
- Use case
- ASR testing· Paired fixture
json
230 BDTMF Transcript: 13579 (JSON)
Ground-truth digit transcript for the DTMF sequence 13579.
- File
- JSON · Dtmf
- Use case
- ASR testing· Paired fixture
json
233 BDTMF Transcript: 567890 (JSON)
Ground-truth digit transcript for the DTMF sequence 567890.
- File
- JSON · Dtmf
- Use case
- ASR testing· Paired fixture
Loudness Ladder
wav
31.3 KBLoudness Ladder: Loud (440 Hz)
1 s 440 Hz tone at linear amplitude 0.55 (~-5.2 dBFS peak): rung on the Wave E loudness ladder.
- File
- WAV · Loudness Ladder · 1 s
- Use case
- Loudness / LUFSAudio analysis· Conversion set
wav
31.3 KBLoudness Ladder: Moderate (440 Hz)
1 s 440 Hz tone at linear amplitude 0.15 (~-16.5 dBFS peak): rung on the Wave E loudness ladder.
- File
- WAV · Loudness Ladder · 1 s
- Use case
- Loudness / LUFSAudio analysis· Conversion set
wav
31.3 KBLoudness Ladder: Near Fs (440 Hz)
1 s 440 Hz tone at linear amplitude 0.95 (~-0.45 dBFS peak): rung on the Wave E loudness ladder.
- File
- WAV · Loudness Ladder · 1 s
- Use case
- Loudness / LUFSAudio analysis· Conversion set
wav
31.3 KBLoudness Ladder: Nominal (440 Hz)
1 s 440 Hz tone at linear amplitude 0.35 (~-9.1 dBFS peak): rung on the Wave E loudness ladder.
- File
- WAV · Loudness Ladder · 1 s
- Use case
- Loudness / LUFSAudio analysis· Conversion set
wav
31.3 KBLoudness Ladder: Quiet (440 Hz)
1 s 440 Hz tone at linear amplitude 0.05 (~-26.0 dBFS peak): rung on the Wave E loudness ladder.
- File
- WAV · Loudness Ladder · 1 s
- Use case
- Loudness / LUFSAudio analysis· Conversion set
wav
31.3 KBLoudness Ladder: Silent (440 Hz)
1 s 440 Hz tone at linear amplitude 0.0 (digital silence): rung on the Wave E loudness ladder.
- File
- WAV · Loudness Ladder · 1 s
- Use case
- Loudness / LUFSAudio analysis· Conversion set
wav
31.3 KBLoudness Ladder: Very Loud (440 Hz)
1 s 440 Hz tone at linear amplitude 0.75 (~-2.5 dBFS peak): rung on the Wave E loudness ladder.
- File
- WAV · Loudness Ladder · 1 s
- Use case
- Loudness / LUFSAudio analysis· Conversion set
wav
31.3 KBLoudness Ladder: Very Quiet (440 Hz)
1 s 440 Hz tone at linear amplitude 0.02 (~-34.0 dBFS peak): rung on the Wave E loudness ladder.
- File
- WAV · Loudness Ladder · 1 s
- Use case
- Loudness / LUFSAudio analysis· Conversion set
P8 Convert
dts
66.5 KBConvert v2 DTS Coherent Acoustics Reference Tone
Real DTS Coherent Acoustics audio payload encoded from the same deterministic 0.7-second mono reference tone for legacy import coverage. Stable P8 artifact p8-convert-legacy-dts.
- File
- DTS · P8 Convert
- Use case
- Conversion testingAudio analysis
mmf
3.1 KBConvert v2 Yamaha SMAF Reference Tone
Real Yamaha SMAF audio payload encoded from the same deterministic 0.7-second mono reference tone for legacy import coverage. Stable P8 artifact p8-convert-legacy-mmf.
- File
- MMF · P8 Convert
- Use case
- Conversion testingAudio analysis
Peak Clip
wav
25 KBPeak-Clip Pair 01: Clean (220 Hz)
Clean 220 Hz tone (amp 0.45): reference twin for the hard-clipped pair.
- File
- WAV · Peak Clip · 16000
- Use case
- Loudness / LUFSAudio analysis· Paired fixture
wav
25 KBPeak-Clip Pair 01: Clipped (220 Hz)
Hard-clipped twin of the 220 Hz tone (×3.2 then clip to ±1) for peak/limiter tests.
- File
- WAV · Peak Clip · 16000
- Use case
- Loudness / LUFSAudio analysis· Paired fixture
wav
25 KBPeak-Clip Pair 02: Clean (440 Hz)
Clean 440 Hz tone (amp 0.45): reference twin for the hard-clipped pair.
- File
- WAV · Peak Clip · 16000
- Use case
- Loudness / LUFSAudio analysis· Paired fixture
wav
25 KBPeak-Clip Pair 02: Clipped (440 Hz)
Hard-clipped twin of the 440 Hz tone (×3.2 then clip to ±1) for peak/limiter tests.
- File
- WAV · Peak Clip · 16000
- Use case
- Loudness / LUFSAudio analysis· Paired fixture
wav
25 KBPeak-Clip Pair 03: Clean (880 Hz)
Clean 880 Hz tone (amp 0.45): reference twin for the hard-clipped pair.
- File
- WAV · Peak Clip · 16000
- Use case
- Loudness / LUFSAudio analysis· Paired fixture
wav
25 KBPeak-Clip Pair 03: Clipped (880 Hz)
Hard-clipped twin of the 880 Hz tone (×3.2 then clip to ±1) for peak/limiter tests.
- File
- WAV · Peak Clip · 16000
- Use case
- Loudness / LUFSAudio analysis· Paired fixture
wav
25 KBPeak-Clip Pair 04: Clean (660 Hz)
Clean 660 Hz tone (amp 0.45): reference twin for the hard-clipped pair.
- File
- WAV · Peak Clip · 16000
- Use case
- Loudness / LUFSAudio analysis· Paired fixture
wav
25 KBPeak-Clip Pair 04: Clipped (660 Hz)
Hard-clipped twin of the 660 Hz tone (×3.2 then clip to ±1) for peak/limiter tests.
- File
- WAV · Peak Clip · 16000
- Use case
- Loudness / LUFSAudio analysis· Paired fixture
wav
25 KBPeak-Clip Pair 05: Clean (330 Hz)
Clean 330 Hz tone (amp 0.45): reference twin for the hard-clipped pair.
- File
- WAV · Peak Clip · 16000
- Use case
- Loudness / LUFSAudio analysis· Paired fixture
wav
25 KBPeak-Clip Pair 05: Clipped (330 Hz)
Hard-clipped twin of the 330 Hz tone (×3.2 then clip to ±1) for peak/limiter tests.
- File
- WAV · Peak Clip · 16000
- Use case
- Loudness / LUFSAudio analysis· Paired fixture
wav
25 KBPeak-Clip Pair 06: Clean (550 Hz)
Clean 550 Hz tone (amp 0.45): reference twin for the hard-clipped pair.
- File
- WAV · Peak Clip · 16000
- Use case
- Loudness / LUFSAudio analysis· Paired fixture
wav
25 KBPeak-Clip Pair 06: Clipped (550 Hz)
Hard-clipped twin of the 550 Hz tone (×3.2 then clip to ±1) for peak/limiter tests.
- File
- WAV · Peak Clip · 16000
- Use case
- Loudness / LUFSAudio analysis· Paired fixture
Reference Tones
wav
258.4 KB1 kHz Sine Tone
A pure 1000 Hz sine tone, 3 seconds, 16-bit 44.1 kHz mono: a clean reference signal for level metering, spectrum analysis, and waveform rendering.
- File
- WAV · Reference Tones · 3 s
- Use case
- Audio analysis
wav
258.4 KB220 Hz Sine Tone
A pure 220 Hz sine tone, 3 seconds, 16-bit 44.1 kHz mono: a clean reference signal for level metering, spectrum analysis, and waveform rendering.
- File
- WAV · Reference Tones · 3 s
- Use case
- Audio analysis
wav
258.4 KB440 Hz Sine Tone (A4)
A pure 440 Hz sine tone, 3 seconds, 16-bit 44.1 kHz mono: a clean reference signal for level metering, spectrum analysis, and waveform rendering.
- File
- WAV · Reference Tones · 3 s
- Use case
- Audio analysis
wav
430.7 KBFrequency Sweep 20 Hz–20 kHz
A 5-second linear sweep from 20 Hz to 20 kHz across the full audible range, for testing frequency response, spectrograms, and playback fidelity.
- File
- WAV · Reference Tones · 5 s
- Use case
- Audio analysis
wav
258.4 KBPink Noise
Three seconds of pink noise with a 1/f power spectrum: the standard reference for loudness and room-calibration testing.
- File
- WAV · Reference Tones · 3 s
- Use case
- Audio analysis
wav
258.4 KBWhite Noise
Three seconds of white noise with a flat power spectrum: a reference for testing noise handling, gating, and spectral tools.
- File
- WAV · Reference Tones · 3 s
- Use case
- Audio analysis
Sample Rate Ladder
wav
62.5 KB1 kHz Tone @ 16000 Hz Sample Rate (WAV)
A 1 kHz sine tone sampled at 16000 Hz: part of a sample-rate ladder (8/16/44.1/48/96 kHz) of the identical tone, for testing resampling, sample-rate detection, and aliasing.
- File
- WAV · Sample Rate Ladder · 2 s
- Use case
- Audio analysisConversion testing· Conversion set
wav
172.3 KB1 kHz Tone @ 44100 Hz Sample Rate (WAV)
A 1 kHz sine tone sampled at 44100 Hz: part of a sample-rate ladder (8/16/44.1/48/96 kHz) of the identical tone, for testing resampling, sample-rate detection, and aliasing.
- File
- WAV · Sample Rate Ladder · 2 s
- Use case
- Audio analysisConversion testing· Conversion set
wav
187.5 KB1 kHz Tone @ 48000 Hz Sample Rate (WAV)
A 1 kHz sine tone sampled at 48000 Hz: part of a sample-rate ladder (8/16/44.1/48/96 kHz) of the identical tone, for testing resampling, sample-rate detection, and aliasing.
- File
- WAV · Sample Rate Ladder · 2 s
- Use case
- Audio analysisConversion testing· Conversion set
wav
31.3 KB1 kHz Tone @ 8000 Hz Sample Rate (WAV)
A 1 kHz sine tone sampled at 8000 Hz: part of a sample-rate ladder (8/16/44.1/48/96 kHz) of the identical tone, for testing resampling, sample-rate detection, and aliasing.
- File
- WAV · Sample Rate Ladder · 2 s
- Use case
- Audio analysisConversion testing· Conversion set
wav
375 KB1 kHz Tone @ 96000 Hz Sample Rate (WAV)
A 1 kHz sine tone sampled at 96000 Hz: part of a sample-rate ladder (8/16/44.1/48/96 kHz) of the identical tone, for testing resampling, sample-rate detection, and aliasing.
- File
- WAV · Sample Rate Ladder · 2 s
- Use case
- Audio analysisConversion testing· Conversion set
Sample Rate Set
wav
18.8 KBSample-Rate Set: 440 Hz @ 16000 Hz
Same 440 Hz tone rendered at 16000 Hz sample rate: twin in the 8k/16k/44.1k set.
- File
- WAV · Sample Rate Set · 0.6 s
- Use case
- Audio analysisConversion testing· Conversion set
wav
51.7 KBSample-Rate Set: 440 Hz @ 44100 Hz
Same 440 Hz tone rendered at 44100 Hz sample rate: twin in the 8k/16k/44.1k set.
- File
- WAV · Sample Rate Set · 0.6 s
- Use case
- Audio analysisConversion testing· Conversion set
wav
9.4 KBSample-Rate Set: 440 Hz @ 8000 Hz
Same 440 Hz tone rendered at 8000 Hz sample rate: twin in the 8k/16k/44.1k set.
- File
- WAV · Sample Rate Set · 0.6 s
- Use case
- Audio analysisConversion testing· Conversion set
wav
15.7 KBSample-Rate Set: 880 Hz @ 16000 Hz
880 Hz tone at 16000 Hz: second ladder for resampling QA.
- File
- WAV · Sample Rate Set · 0.5 s
- Use case
- Audio analysisConversion testing· Conversion set
wav
43.1 KBSample-Rate Set: 880 Hz @ 44100 Hz
880 Hz tone at 44100 Hz: second ladder for resampling QA.
- File
- WAV · Sample Rate Set · 0.5 s
- Use case
- Audio analysisConversion testing· Conversion set
wav
7.9 KBSample-Rate Set: 880 Hz @ 8000 Hz
880 Hz tone at 8000 Hz: second ladder for resampling QA.
- File
- WAV · Sample Rate Set · 0.5 s
- Use case
- Audio analysisConversion testing· Conversion set
Signals
wav
172.3 KBClipped Tone: intentionally clipped (WAV)
A 440 Hz tone driven past full scale and hard-clipped so its peaks are flat-topped: an intentionally clipped signal for testing clip and true-peak detection, distortion metering, and de-clipping tools.
- File
- WAV · Signals · 2 s
- Use case
- Audio analysisError handling
wav
172.3 KBDigital Silence (2s WAV)
Two seconds of true digital silence (all-zero samples): a fixture for testing silence detection, auto-trim thresholds, and noise-floor handling.
- File
- WAV · Signals · 2 s
- Use case
- Audio analysisAuto-trim testing
wav
172.3 KBMelody: C Major Scale (WAV)
A two-second melody playing an ascending C-major scale (C4 to C5) with per-note envelopes: a simple musical signal for testing pitch detection, onset detection, and playback.
- File
- WAV · Signals · 2 s
- Use case
- Audio analysis
wav
89.6 KBNotification Chime (WAV)
A short four-note notification chime (a C-major arpeggio with a decaying final note): a pleasant, recognisable UI sound for testing playback, short-clip handling, and notification pipelines.
- File
- WAV · Signals · 1 s
- Use case
- Audio analysis
wav
129.2 KBQuiet Audio Needing Normalization
A very quiet modulated tone (~peak 0.04) for testing loudness normalization and gain staging.
- File
- WAV · Signals · 1.5 s
- Use case
- Audio analysis
wav
172.3 KBStereo Channel Imbalance (440 Hz)
A 440 Hz stereo tone with a loud left channel and a near-silent right channel, for balance and mono-mix testing.
- File
- WAV · Signals · 1 s
- Use case
- Audio analysis
wav
86.2 KBTone with Click/Pop Transient
A clean 440 Hz tone interrupted by a short high-amplitude click at 0.4s, for click/pop removal tools.
- File
- WAV · Signals · 1 s
- Use case
- Audio analysisError handling
Silence Middle
wav
46.9 KBSilence-in-Middle: Asym Lead
440 Hz tone with 0.5s of digital silence in the middle (0.2s + silence + 0.8s). Fixture for mid-gap trim / split tools.
- File
- WAV · Silence Middle · 16000
- Use case
- Auto-trim testingAudio analysis· Conversion set
wav
46.9 KBSilence-in-Middle: Asym Trail
440 Hz tone with 0.5s of digital silence in the middle (0.8s + silence + 0.2s). Fixture for mid-gap trim / split tools.
- File
- WAV · Silence Middle · 16000
- Use case
- Auto-trim testingAudio analysis· Conversion set
wav
71.9 KBSilence-in-Middle: Long Gap
440 Hz tone with 1.5s of digital silence in the middle (0.4s + silence + 0.4s). Fixture for mid-gap trim / split tools.
- File
- WAV · Silence Middle · 16000
- Use case
- Auto-trim testingAudio analysis· Conversion set
wav
54.7 KBSilence-in-Middle: Medium Gap
440 Hz tone with 0.75s of digital silence in the middle (0.5s + silence + 0.5s). Fixture for mid-gap trim / split tools.
- File
- WAV · Silence Middle · 16000
- Use case
- Auto-trim testingAudio analysis· Conversion set
wav
32.9 KBSilence-in-Middle: Short Gap
440 Hz tone with 0.25s of digital silence in the middle (0.4s + silence + 0.4s). Fixture for mid-gap trim / split tools.
- File
- WAV · Silence Middle · 16000
- Use case
- Auto-trim testingAudio analysis· Conversion set
wav
24.4 KBSilence-in-Middle: Tiny Gap
440 Hz tone with 0.08s of digital silence in the middle (0.35s + silence + 0.35s). Fixture for mid-gap trim / split tools.
- File
- WAV · Silence Middle · 16000
- Use case
- Auto-trim testingAudio analysis· Conversion set
Silence Trim Set
wav
258.4 KBLeading Silence 1s: 440 Hz Tone
A 440 Hz tone preceded by exactly 1 second of digital silence. A direct fixture for auto-trim tools: the tone should start at 1.000s.
- File
- WAV · Silence Trim Set · 44100 Hz
- Use case
- Auto-trim testingAudio analysis
wav
430.7 KBLeading Silence 3s: 440 Hz Tone
A 440 Hz tone preceded by exactly 3 seconds of digital silence. A direct fixture for auto-trim tools: the tone should start at 3.000s.
- File
- WAV · Silence Trim Set · 44100 Hz
- Use case
- Auto-trim testingAudio analysis
wav
603 KBLeading Silence 5s: 440 Hz Tone
A 440 Hz tone preceded by exactly 5 seconds of digital silence. A direct fixture for auto-trim tools: the tone should start at 5.000s.
- File
- WAV · Silence Trim Set · 44100 Hz
- Use case
- Auto-trim testingAudio analysis
wav
258.4 KBTrailing Silence 1s: 440 Hz Tone
A 440 Hz tone followed by exactly 1 second of digital silence. The tone should end at 2.0s: a direct fixture for testing trailing-silence trimming.
- File
- WAV · Silence Trim Set · 44100 Hz
- Use case
- Auto-trim testingAudio analysis
wav
430.7 KBTrailing Silence 3s: 440 Hz Tone
A 440 Hz tone followed by exactly 3 seconds of digital silence. The tone should end at 2.0s: a direct fixture for testing trailing-silence trimming.
- File
- WAV · Silence Trim Set · 44100 Hz
- Use case
- Auto-trim testingAudio analysis
wav
603 KBTrailing Silence 5s: 440 Hz Tone
A 440 Hz tone followed by exactly 5 seconds of digital silence. The tone should end at 2.0s: a direct fixture for testing trailing-silence trimming.
- File
- WAV · Silence Trim Set · 44100 Hz
- Use case
- Auto-trim testingAudio analysis
Speech Ladders
wav
39.2 KBComplaint call: chain telephone
A 5 second excerpt of this group's take, telephone chain. Band-limited, resampled and companded in one pass: the whole telephone path applied at once rather than one effect at a time. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
145.6 KBComplaint call: clipped
A 5 second excerpt of this group's take, clipping. Samples driven past a fixed threshold and flattened, the damage a hot microphone does and the one no codec can undo. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBComplaint call: codec g722 16k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as adpcm_g722 at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · adpcm_g722
- Use case
- ASR testingAudio analysis+1· Conversion set
gsm
8.1 KBComplaint call: codec gsm 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as gsm at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- GSM · Speech Ladders · gsm
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
79.5 KBComplaint call: codec mp3 128
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
20 KBComplaint call: codec mp3 32
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBComplaint call: codec mulaw 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
opus
14.7 KBComplaint call: codec opus 24
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as opus at 48000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- OPUS · Speech Ladders · opus
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
135 KBComplaint call: dropouts
A 5 second excerpt of this group's take, dropouts. Short spans silenced outright, as a packet-switched call drops them, so a transcriber must decide between a pause and a missing word. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
382.3 KBComplaint call: full spoken take
The complete 13.5 second utterance at 24000 Hz mono FLAC, spoken by the bf_isabella voice at 0.95x rate. The five-second excerpt in this group's ladder is cut from this take, so anything needing real duration uses this file and anything comparing damage uses the ladder. Lossless, so it is the archival reference for the whole group.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis· Conversion set
flac
333 KBComplaint call: noise snr0
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
313.8 KBComplaint call: noise snr10
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
297.4 KBComplaint call: noise snr20
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
145.7 KBComplaint call: plate
A 5 second excerpt of this group's take, reference. The clean five-second excerpt every other file in this ladder is scored against. Delivered as flac at 24000 Hz mono. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
186.3 KBComplaint call: rate 16000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
391.6 KBComplaint call: rate 44100
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 44100 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
97.4 KBComplaint call: rate 8000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
txt
183 BComplaint call: recogniser transcript
What a speech recogniser returned for this take. It normalises: spoken "forty three dollars and eighteen cents" comes back as digits and a currency symbol. Compared with the reference script without normalising first, this disagrees on every number in the utterance and a word error rate computed that way reports a fault that does not exist. That gap is the point of the pair.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
txt
227 BComplaint call: reference script
Exactly what was spoken, which is what the synthesiser was given: numbers, currency and times are spelled out as words because that is how they were said. This is ground truth by construction - it existed before the audio did - and it is deliberately NOT the same string as the recogniser transcript paired with it.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
srt
316 BComplaint call: timed captions
The recogniser's output with timings, 4 cues over the full take. Carries the same normalisation as the plain transcript in this group, so it inherits the same scoring trap, and adds the timing dimension: a cue that starts before the word is spoken is a different defect from a cue with the wrong words in it.
- File
- SRT · Speech Ladders · 4 cues
- Use case
- ASR testingMedia accessibility· Conversion set
wav
39.2 KBCost podcast: chain telephone
A 5 second excerpt of this group's take, telephone chain. Band-limited, resampled and companded in one pass: the whole telephone path applied at once rather than one effect at a time. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
132.2 KBCost podcast: clipped
A 5 second excerpt of this group's take, clipping. Samples driven past a fixed threshold and flattened, the damage a hot microphone does and the one no codec can undo. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBCost podcast: codec g722 16k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as adpcm_g722 at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · adpcm_g722
- Use case
- ASR testingAudio analysis+1· Conversion set
gsm
8.1 KBCost podcast: codec gsm 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as gsm at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- GSM · Speech Ladders · gsm
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
79.5 KBCost podcast: codec mp3 128
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
20 KBCost podcast: codec mp3 32
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBCost podcast: codec mulaw 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
opus
15.3 KBCost podcast: codec opus 24
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as opus at 48000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- OPUS · Speech Ladders · opus
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
122 KBCost podcast: dropouts
A 5 second excerpt of this group's take, dropouts. Short spans silenced outright, as a packet-switched call drops them, so a transcriber must decide between a pause and a missing word. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
418 KBCost podcast: full spoken take
The complete 16.5 second utterance at 24000 Hz mono FLAC, spoken by the am_eric voice at 1.0x rate. The five-second excerpt in this group's ladder is cut from this take, so anything needing real duration uses this file and anything comparing damage uses the ladder. Lossless, so it is the archival reference for the whole group.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis· Conversion set
flac
327.1 KBCost podcast: noise snr0
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
306.2 KBCost podcast: noise snr10
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
288.5 KBCost podcast: noise snr20
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
130.7 KBCost podcast: plate
A 5 second excerpt of this group's take, reference. The clean five-second excerpt every other file in this ladder is scored against. Delivered as flac at 24000 Hz mono. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
186.2 KBCost podcast: rate 16000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
332.9 KBCost podcast: rate 44100
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 44100 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
98.9 KBCost podcast: rate 8000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
txt
313 BCost podcast: recogniser transcript
What a speech recogniser returned for this take. It normalises: spoken "forty three dollars and eighteen cents" comes back as digits and a currency symbol. Compared with the reference script without normalising first, this disagrees on every number in the utterance and a word error rate computed that way reports a fault that does not exist. That gap is the point of the pair.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
txt
312 BCost podcast: reference script
Exactly what was spoken, which is what the synthesiser was given: numbers, currency and times are spelled out as words because that is how they were said. This is ground truth by construction - it existed before the audio did - and it is deliberately NOT the same string as the recogniser transcript paired with it.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
srt
403 BCost podcast: timed captions
The recogniser's output with timings, 4 cues over the full take. Carries the same normalisation as the plain transcript in this group, so it inherits the same scoring trap, and adds the timing dimension: a cue that starts before the word is spoken is a different defect from a cue with the wrong words in it.
- File
- SRT · Speech Ladders · 4 cues
- Use case
- ASR testingMedia accessibility· Conversion set
wav
39.2 KBDelivery driver: chain telephone
A 5 second excerpt of this group's take, telephone chain. Band-limited, resampled and companded in one pass: the whole telephone path applied at once rather than one effect at a time. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
148.8 KBDelivery driver: clipped
A 5 second excerpt of this group's take, clipping. Samples driven past a fixed threshold and flattened, the damage a hot microphone does and the one no codec can undo. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBDelivery driver: codec g722 16k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as adpcm_g722 at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · adpcm_g722
- Use case
- ASR testingAudio analysis+1· Conversion set
gsm
8.1 KBDelivery driver: codec gsm 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as gsm at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- GSM · Speech Ladders · gsm
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
79.5 KBDelivery driver: codec mp3 128
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
20 KBDelivery driver: codec mp3 32
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBDelivery driver: codec mulaw 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
opus
15.1 KBDelivery driver: codec opus 24
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as opus at 48000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- OPUS · Speech Ladders · opus
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
138 KBDelivery driver: dropouts
A 5 second excerpt of this group's take, dropouts. Short spans silenced outright, as a packet-switched call drops them, so a transcriber must decide between a pause and a missing word. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
361.6 KBDelivery driver: full spoken take
The complete 13.0 second utterance at 24000 Hz mono FLAC, spoken by the am_adam voice at 1.05x rate. The five-second excerpt in this group's ladder is cut from this take, so anything needing real duration uses this file and anything comparing damage uses the ladder. Lossless, so it is the archival reference for the whole group.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis· Conversion set
flac
336 KBDelivery driver: noise snr0
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
316.4 KBDelivery driver: noise snr10
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
300.2 KBDelivery driver: noise snr20
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
147.8 KBDelivery driver: plate
A 5 second excerpt of this group's take, reference. The clean five-second excerpt every other file in this ladder is scored against. Delivered as flac at 24000 Hz mono. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
190.9 KBDelivery driver: rate 16000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
389.9 KBDelivery driver: rate 44100
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 44100 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
100.8 KBDelivery driver: rate 8000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
txt
241 BDelivery driver: recogniser transcript
What a speech recogniser returned for this take. It normalises: spoken "forty three dollars and eighteen cents" comes back as digits and a currency symbol. Compared with the reference script without normalising first, this disagrees on every number in the utterance and a word error rate computed that way reports a fault that does not exist. That gap is the point of the pair.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
txt
241 BDelivery driver: reference script
Exactly what was spoken, which is what the synthesiser was given: numbers, currency and times are spelled out as words because that is how they were said. This is ground truth by construction - it existed before the audio did - and it is deliberately NOT the same string as the recogniser transcript paired with it.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
srt
438 BDelivery driver: timed captions
The recogniser's output with timings, 6 cues over the full take. Carries the same normalisation as the plain transcript in this group, so it inherits the same scoring trap, and adds the timing dimension: a cue that starts before the word is spoken is a different defect from a cue with the wrong words in it.
- File
- SRT · Speech Ladders · 6 cues
- Use case
- ASR testingMedia accessibility· Conversion set
wav
39.2 KBHaccp training: chain telephone
A 5 second excerpt of this group's take, telephone chain. Band-limited, resampled and companded in one pass: the whole telephone path applied at once rather than one effect at a time. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
146.1 KBHaccp training: clipped
A 5 second excerpt of this group's take, clipping. Samples driven past a fixed threshold and flattened, the damage a hot microphone does and the one no codec can undo. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBHaccp training: codec g722 16k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as adpcm_g722 at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · adpcm_g722
- Use case
- ASR testingAudio analysis+1· Conversion set
gsm
8.1 KBHaccp training: codec gsm 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as gsm at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- GSM · Speech Ladders · gsm
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
79.5 KBHaccp training: codec mp3 128
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
20 KBHaccp training: codec mp3 32
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBHaccp training: codec mulaw 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
opus
15.4 KBHaccp training: codec opus 24
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as opus at 48000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- OPUS · Speech Ladders · opus
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
136.8 KBHaccp training: dropouts
A 5 second excerpt of this group's take, dropouts. Short spans silenced outright, as a packet-switched call drops them, so a transcriber must decide between a pause and a missing word. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
506.2 KBHaccp training: full spoken take
The complete 19.2 second utterance at 24000 Hz mono FLAC, spoken by the bm_george voice at 0.9x rate. The five-second excerpt in this group's ladder is cut from this take, so anything needing real duration uses this file and anything comparing damage uses the ladder. Lossless, so it is the archival reference for the whole group.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis· Conversion set
flac
329.6 KBHaccp training: noise snr0
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
309.4 KBHaccp training: noise snr10
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
292.8 KBHaccp training: noise snr20
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
146 KBHaccp training: plate
A 5 second excerpt of this group's take, reference. The clean five-second excerpt every other file in this ladder is scored against. Delivered as flac at 24000 Hz mono. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
192.2 KBHaccp training: rate 16000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
359.5 KBHaccp training: rate 44100
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 44100 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
99.5 KBHaccp training: rate 8000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
txt
270 BHaccp training: recogniser transcript
What a speech recogniser returned for this take. It normalises: spoken "forty three dollars and eighteen cents" comes back as digits and a currency symbol. Compared with the reference script without normalising first, this disagrees on every number in the utterance and a word error rate computed that way reports a fault that does not exist. That gap is the point of the pair.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
txt
289 BHaccp training: reference script
Exactly what was spoken, which is what the synthesiser was given: numbers, currency and times are spelled out as words because that is how they were said. This is ground truth by construction - it existed before the audio did - and it is deliberately NOT the same string as the recogniser transcript paired with it.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
srt
401 BHaccp training: timed captions
The recogniser's output with timings, 4 cues over the full take. Carries the same normalisation as the plain transcript in this group, so it inherits the same scoring trap, and adds the timing dimension: a cue that starts before the word is spoken is a different defect from a cue with the wrong words in it.
- File
- SRT · Speech Ladders · 4 cues
- Use case
- ASR testingMedia accessibility· Conversion set
wav
39.2 KBHomophone stress: chain telephone
A 5 second excerpt of this group's take, telephone chain. Band-limited, resampled and companded in one pass: the whole telephone path applied at once rather than one effect at a time. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
129.5 KBHomophone stress: clipped
A 5 second excerpt of this group's take, clipping. Samples driven past a fixed threshold and flattened, the damage a hot microphone does and the one no codec can undo. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBHomophone stress: codec g722 16k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as adpcm_g722 at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · adpcm_g722
- Use case
- ASR testingAudio analysis+1· Conversion set
gsm
8.1 KBHomophone stress: codec gsm 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as gsm at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- GSM · Speech Ladders · gsm
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
79.5 KBHomophone stress: codec mp3 128
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
20 KBHomophone stress: codec mp3 32
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBHomophone stress: codec mulaw 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
opus
14.7 KBHomophone stress: codec opus 24
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as opus at 48000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- OPUS · Speech Ladders · opus
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
119.5 KBHomophone stress: dropouts
A 5 second excerpt of this group's take, dropouts. Short spans silenced outright, as a packet-switched call drops them, so a transcriber must decide between a pause and a missing word. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
327.4 KBHomophone stress: full spoken take
The complete 13.3 second utterance at 24000 Hz mono FLAC, spoken by the bm_lewis voice at 1.0x rate. The five-second excerpt in this group's ladder is cut from this take, so anything needing real duration uses this file and anything comparing damage uses the ladder. Lossless, so it is the archival reference for the whole group.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis· Conversion set
flac
325.2 KBHomophone stress: noise snr0
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
303.8 KBHomophone stress: noise snr10
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
286.6 KBHomophone stress: noise snr20
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
129.4 KBHomophone stress: plate
A 5 second excerpt of this group's take, reference. The clean five-second excerpt every other file in this ladder is scored against. Delivered as flac at 24000 Hz mono. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
172.4 KBHomophone stress: rate 16000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
352.1 KBHomophone stress: rate 44100
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 44100 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
92.6 KBHomophone stress: rate 8000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
txt
215 BHomophone stress: recogniser transcript
What a speech recogniser returned for this take. It normalises: spoken "forty three dollars and eighteen cents" comes back as digits and a currency symbol. Compared with the reference script without normalising first, this disagrees on every number in the utterance and a word error rate computed that way reports a fault that does not exist. That gap is the point of the pair.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
txt
216 BHomophone stress: reference script
Exactly what was spoken, which is what the synthesiser was given: numbers, currency and times are spelled out as words because that is how they were said. This is ground truth by construction - it existed before the audio did - and it is deliberately NOT the same string as the recogniser transcript paired with it.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
srt
379 BHomophone stress: timed captions
The recogniser's output with timings, 5 cues over the full take. Carries the same normalisation as the plain transcript in this group, so it inherits the same scoring trap, and adds the timing dimension: a cue that starts before the word is spoken is a different defect from a cue with the wrong words in it.
- File
- SRT · Speech Ladders · 5 cues
- Use case
- ASR testingMedia accessibility· Conversion set
wav
39.2 KBInvoice dispute: chain telephone
A 5 second excerpt of this group's take, telephone chain. Band-limited, resampled and companded in one pass: the whole telephone path applied at once rather than one effect at a time. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
137.2 KBInvoice dispute: clipped
A 5 second excerpt of this group's take, clipping. Samples driven past a fixed threshold and flattened, the damage a hot microphone does and the one no codec can undo. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBInvoice dispute: codec g722 16k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as adpcm_g722 at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · adpcm_g722
- Use case
- ASR testingAudio analysis+1· Conversion set
gsm
8.1 KBInvoice dispute: codec gsm 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as gsm at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- GSM · Speech Ladders · gsm
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
79.5 KBInvoice dispute: codec mp3 128
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
20 KBInvoice dispute: codec mp3 32
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBInvoice dispute: codec mulaw 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
opus
15.1 KBInvoice dispute: codec opus 24
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as opus at 48000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- OPUS · Speech Ladders · opus
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
127.8 KBInvoice dispute: dropouts
A 5 second excerpt of this group's take, dropouts. Short spans silenced outright, as a packet-switched call drops them, so a transcriber must decide between a pause and a missing word. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
537.8 KBInvoice dispute: full spoken take
The complete 20.2 second utterance at 24000 Hz mono FLAC, spoken by the am_michael voice at 0.95x rate. The five-second excerpt in this group's ladder is cut from this take, so anything needing real duration uses this file and anything comparing damage uses the ladder. Lossless, so it is the archival reference for the whole group.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis· Conversion set
flac
323.5 KBInvoice dispute: noise snr0
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
301.9 KBInvoice dispute: noise snr10
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
284.6 KBInvoice dispute: noise snr20
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
137 KBInvoice dispute: plate
A 5 second excerpt of this group's take, reference. The clean five-second excerpt every other file in this ladder is scored against. Delivered as flac at 24000 Hz mono. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
181.2 KBInvoice dispute: rate 16000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
367.4 KBInvoice dispute: rate 44100
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 44100 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
95.7 KBInvoice dispute: rate 8000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
txt
233 BInvoice dispute: recogniser transcript
What a speech recogniser returned for this take. It normalises: spoken "forty three dollars and eighteen cents" comes back as digits and a currency symbol. Compared with the reference script without normalising first, this disagrees on every number in the utterance and a word error rate computed that way reports a fault that does not exist. That gap is the point of the pair.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
txt
304 BInvoice dispute: reference script
Exactly what was spoken, which is what the synthesiser was given: numbers, currency and times are spelled out as words because that is how they were said. This is ground truth by construction - it existed before the audio did - and it is deliberately NOT the same string as the recogniser transcript paired with it.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
srt
367 BInvoice dispute: timed captions
The recogniser's output with timings, 4 cues over the full take. Carries the same normalisation as the plain transcript in this group, so it inherits the same scoring trap, and adds the timing dimension: a cue that starts before the word is spoken is a different defect from a cue with the wrong words in it.
- File
- SRT · Speech Ladders · 4 cues
- Use case
- ASR testingMedia accessibility· Conversion set
wav
39.2 KBIvr menu: chain telephone
A 5 second excerpt of this group's take, telephone chain. Band-limited, resampled and companded in one pass: the whole telephone path applied at once rather than one effect at a time. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
138.7 KBIvr menu: clipped
A 5 second excerpt of this group's take, clipping. Samples driven past a fixed threshold and flattened, the damage a hot microphone does and the one no codec can undo. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBIvr menu: codec g722 16k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as adpcm_g722 at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · adpcm_g722
- Use case
- ASR testingAudio analysis+1· Conversion set
gsm
8.1 KBIvr menu: codec gsm 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as gsm at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- GSM · Speech Ladders · gsm
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
79.5 KBIvr menu: codec mp3 128
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
20 KBIvr menu: codec mp3 32
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBIvr menu: codec mulaw 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
opus
15 KBIvr menu: codec opus 24
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as opus at 48000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- OPUS · Speech Ladders · opus
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
127.1 KBIvr menu: dropouts
A 5 second excerpt of this group's take, dropouts. Short spans silenced outright, as a packet-switched call drops them, so a transcriber must decide between a pause and a missing word. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
323.2 KBIvr menu: full spoken take
The complete 11.9 second utterance at 24000 Hz mono FLAC, spoken by the af_heart voice at 1.0x rate. The five-second excerpt in this group's ladder is cut from this take, so anything needing real duration uses this file and anything comparing damage uses the ladder. Lossless, so it is the archival reference for the whole group.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis· Conversion set
flac
327.3 KBIvr menu: noise snr0
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
306.5 KBIvr menu: noise snr10
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
288.8 KBIvr menu: noise snr20
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
138.2 KBIvr menu: plate
A 5 second excerpt of this group's take, reference. The clean five-second excerpt every other file in this ladder is scored against. Delivered as flac at 24000 Hz mono. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
181.7 KBIvr menu: rate 16000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
376 KBIvr menu: rate 44100
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 44100 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
96.7 KBIvr menu: rate 8000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
txt
191 BIvr menu: recogniser transcript
What a speech recogniser returned for this take. It normalises: spoken "forty three dollars and eighteen cents" comes back as digits and a currency symbol. Compared with the reference script without normalising first, this disagrees on every number in the utterance and a word error rate computed that way reports a fault that does not exist. That gap is the point of the pair.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
txt
192 BIvr menu: reference script
Exactly what was spoken, which is what the synthesiser was given: numbers, currency and times are spelled out as words because that is how they were said. This is ground truth by construction - it existed before the audio did - and it is deliberately NOT the same string as the recogniser transcript paired with it.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
srt
289 BIvr menu: timed captions
The recogniser's output with timings, 3 cues over the full take. Carries the same normalisation as the plain transcript in this group, so it inherits the same scoring trap, and adds the timing dimension: a cue that starts before the word is spoken is a different defect from a cue with the wrong words in it.
- File
- SRT · Speech Ladders · 3 cues
- Use case
- ASR testingMedia accessibility· Conversion set
wav
39.2 KBMenu narration: chain telephone
A 5 second excerpt of this group's take, telephone chain. Band-limited, resampled and companded in one pass: the whole telephone path applied at once rather than one effect at a time. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
148.1 KBMenu narration: clipped
A 5 second excerpt of this group's take, clipping. Samples driven past a fixed threshold and flattened, the damage a hot microphone does and the one no codec can undo. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBMenu narration: codec g722 16k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as adpcm_g722 at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · adpcm_g722
- Use case
- ASR testingAudio analysis+1· Conversion set
gsm
8.1 KBMenu narration: codec gsm 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as gsm at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- GSM · Speech Ladders · gsm
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
79.5 KBMenu narration: codec mp3 128
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
20 KBMenu narration: codec mp3 32
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBMenu narration: codec mulaw 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
opus
15.1 KBMenu narration: codec opus 24
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as opus at 48000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- OPUS · Speech Ladders · opus
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
139.1 KBMenu narration: dropouts
A 5 second excerpt of this group's take, dropouts. Short spans silenced outright, as a packet-switched call drops them, so a transcriber must decide between a pause and a missing word. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
425.9 KBMenu narration: full spoken take
The complete 15.6 second utterance at 24000 Hz mono FLAC, spoken by the af_bella voice at 0.95x rate. The five-second excerpt in this group's ladder is cut from this take, so anything needing real duration uses this file and anything comparing damage uses the ladder. Lossless, so it is the archival reference for the whole group.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis· Conversion set
flac
329.4 KBMenu narration: noise snr0
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
309.2 KBMenu narration: noise snr10
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
292.6 KBMenu narration: noise snr20
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
148.1 KBMenu narration: plate
A 5 second excerpt of this group's take, reference. The clean five-second excerpt every other file in this ladder is scored against. Delivered as flac at 24000 Hz mono. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
185.2 KBMenu narration: rate 16000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
399.9 KBMenu narration: rate 44100
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 44100 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
94.9 KBMenu narration: rate 8000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
txt
211 BMenu narration: recogniser transcript
What a speech recogniser returned for this take. It normalises: spoken "forty three dollars and eighteen cents" comes back as digits and a currency symbol. Compared with the reference script without normalising first, this disagrees on every number in the utterance and a word error rate computed that way reports a fault that does not exist. That gap is the point of the pair.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
txt
226 BMenu narration: reference script
Exactly what was spoken, which is what the synthesiser was given: numbers, currency and times are spelled out as words because that is how they were said. This is ground truth by construction - it existed before the audio did - and it is deliberately NOT the same string as the recogniser transcript paired with it.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
srt
342 BMenu narration: timed captions
The recogniser's output with timings, 4 cues over the full take. Carries the same normalisation as the plain transcript in this group, so it inherits the same scoring trap, and adds the timing dimension: a cue that starts before the word is spoken is a different defect from a cue with the wrong words in it.
- File
- SRT · Speech Ladders · 4 cues
- Use case
- ASR testingMedia accessibility· Conversion set
wav
39.2 KBNumbers stress: chain telephone
A 5 second excerpt of this group's take, telephone chain. Band-limited, resampled and companded in one pass: the whole telephone path applied at once rather than one effect at a time. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
139.2 KBNumbers stress: clipped
A 5 second excerpt of this group's take, clipping. Samples driven past a fixed threshold and flattened, the damage a hot microphone does and the one no codec can undo. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBNumbers stress: codec g722 16k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as adpcm_g722 at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · adpcm_g722
- Use case
- ASR testingAudio analysis+1· Conversion set
gsm
8.1 KBNumbers stress: codec gsm 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as gsm at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- GSM · Speech Ladders · gsm
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
79.5 KBNumbers stress: codec mp3 128
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
20 KBNumbers stress: codec mp3 32
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBNumbers stress: codec mulaw 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
opus
15.2 KBNumbers stress: codec opus 24
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as opus at 48000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- OPUS · Speech Ladders · opus
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
129 KBNumbers stress: dropouts
A 5 second excerpt of this group's take, dropouts. Short spans silenced outright, as a packet-switched call drops them, so a transcriber must decide between a pause and a missing word. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
485.9 KBNumbers stress: full spoken take
The complete 18.5 second utterance at 24000 Hz mono FLAC, spoken by the am_liam voice at 1.0x rate. The five-second excerpt in this group's ladder is cut from this take, so anything needing real duration uses this file and anything comparing damage uses the ladder. Lossless, so it is the archival reference for the whole group.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis· Conversion set
flac
327.4 KBNumbers stress: noise snr0
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
307.6 KBNumbers stress: noise snr10
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
291.4 KBNumbers stress: noise snr20
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
137.9 KBNumbers stress: plate
A 5 second excerpt of this group's take, reference. The clean five-second excerpt every other file in this ladder is scored against. Delivered as flac at 24000 Hz mono. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
187.3 KBNumbers stress: rate 16000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
346.6 KBNumbers stress: rate 44100
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 44100 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
98.1 KBNumbers stress: rate 8000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
txt
174 BNumbers stress: recogniser transcript
What a speech recogniser returned for this take. It normalises: spoken "forty three dollars and eighteen cents" comes back as digits and a currency symbol. Compared with the reference script without normalising first, this disagrees on every number in the utterance and a word error rate computed that way reports a fault that does not exist. That gap is the point of the pair.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
txt
366 BNumbers stress: reference script
Exactly what was spoken, which is what the synthesiser was given: numbers, currency and times are spelled out as words because that is how they were said. This is ground truth by construction - it existed before the audio did - and it is deliberately NOT the same string as the recogniser transcript paired with it.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
srt
312 BNumbers stress: timed captions
The recogniser's output with timings, 4 cues over the full take. Carries the same normalisation as the plain transcript in this group, so it inherits the same scoring trap, and adds the timing dimension: a cue that starts before the word is spoken is a different defect from a cue with the wrong words in it.
- File
- SRT · Speech Ladders · 4 cues
- Use case
- ASR testingMedia accessibility· Conversion set
wav
39.2 KBReservation voicemail: chain telephone
A 5 second excerpt of this group's take, telephone chain. Band-limited, resampled and companded in one pass: the whole telephone path applied at once rather than one effect at a time. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
151.7 KBReservation voicemail: clipped
A 5 second excerpt of this group's take, clipping. Samples driven past a fixed threshold and flattened, the damage a hot microphone does and the one no codec can undo. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBReservation voicemail: codec g722 16k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as adpcm_g722 at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · adpcm_g722
- Use case
- ASR testingAudio analysis+1· Conversion set
gsm
8.1 KBReservation voicemail: codec gsm 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as gsm at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- GSM · Speech Ladders · gsm
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
79.5 KBReservation voicemail: codec mp3 128
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
20 KBReservation voicemail: codec mp3 32
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBReservation voicemail: codec mulaw 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
opus
14.8 KBReservation voicemail: codec opus 24
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as opus at 48000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- OPUS · Speech Ladders · opus
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
141.9 KBReservation voicemail: dropouts
A 5 second excerpt of this group's take, dropouts. Short spans silenced outright, as a packet-switched call drops them, so a transcriber must decide between a pause and a missing word. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
455.7 KBReservation voicemail: full spoken take
The complete 15.6 second utterance at 24000 Hz mono FLAC, spoken by the bf_emma voice at 1.0x rate. The five-second excerpt in this group's ladder is cut from this take, so anything needing real duration uses this file and anything comparing damage uses the ladder. Lossless, so it is the archival reference for the whole group.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis· Conversion set
flac
329.3 KBReservation voicemail: noise snr0
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
310.4 KBReservation voicemail: noise snr10
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
295.9 KBReservation voicemail: noise snr20
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
151.9 KBReservation voicemail: plate
A 5 second excerpt of this group's take, reference. The clean five-second excerpt every other file in this ladder is scored against. Delivered as flac at 24000 Hz mono. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
187 KBReservation voicemail: rate 16000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
400.8 KBReservation voicemail: rate 44100
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 44100 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
96.9 KBReservation voicemail: rate 8000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
txt
214 BReservation voicemail: recogniser transcript
What a speech recogniser returned for this take. It normalises: spoken "forty three dollars and eighteen cents" comes back as digits and a currency symbol. Compared with the reference script without normalising first, this disagrees on every number in the utterance and a word error rate computed that way reports a fault that does not exist. That gap is the point of the pair.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
txt
262 BReservation voicemail: reference script
Exactly what was spoken, which is what the synthesiser was given: numbers, currency and times are spelled out as words because that is how they were said. This is ground truth by construction - it existed before the audio did - and it is deliberately NOT the same string as the recogniser transcript paired with it.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
srt
379 BReservation voicemail: timed captions
The recogniser's output with timings, 5 cues over the full take. Carries the same normalisation as the plain transcript in this group, so it inherits the same scoring trap, and adds the timing dimension: a cue that starts before the word is spoken is a different defect from a cue with the wrong words in it.
- File
- SRT · Speech Ladders · 5 cues
- Use case
- ASR testingMedia accessibility· Conversion set
wav
39.2 KBStaff briefing: chain telephone
A 5 second excerpt of this group's take, telephone chain. Band-limited, resampled and companded in one pass: the whole telephone path applied at once rather than one effect at a time. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
135.1 KBStaff briefing: clipped
A 5 second excerpt of this group's take, clipping. Samples driven past a fixed threshold and flattened, the damage a hot microphone does and the one no codec can undo. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBStaff briefing: codec g722 16k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as adpcm_g722 at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · adpcm_g722
- Use case
- ASR testingAudio analysis+1· Conversion set
gsm
8.1 KBStaff briefing: codec gsm 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as gsm at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- GSM · Speech Ladders · gsm
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
79.5 KBStaff briefing: codec mp3 128
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
20 KBStaff briefing: codec mp3 32
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBStaff briefing: codec mulaw 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
opus
14.1 KBStaff briefing: codec opus 24
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as opus at 48000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- OPUS · Speech Ladders · opus
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
124.8 KBStaff briefing: dropouts
A 5 second excerpt of this group's take, dropouts. Short spans silenced outright, as a packet-switched call drops them, so a transcriber must decide between a pause and a missing word. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
622.8 KBStaff briefing: full spoken take
The complete 26.2 second utterance at 24000 Hz mono FLAC, spoken by the af_nicole voice at 1.0x rate. The five-second excerpt in this group's ladder is cut from this take, so anything needing real duration uses this file and anything comparing damage uses the ladder. Lossless, so it is the archival reference for the whole group.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis· Conversion set
flac
326.3 KBStaff briefing: noise snr0
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
305.3 KBStaff briefing: noise snr10
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
288.7 KBStaff briefing: noise snr20
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
135.5 KBStaff briefing: plate
A 5 second excerpt of this group's take, reference. The clean five-second excerpt every other file in this ladder is scored against. Delivered as flac at 24000 Hz mono. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
176.2 KBStaff briefing: rate 16000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
372.7 KBStaff briefing: rate 44100
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 44100 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
91.9 KBStaff briefing: rate 8000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
txt
252 BStaff briefing: recogniser transcript
What a speech recogniser returned for this take. It normalises: spoken "forty three dollars and eighteen cents" comes back as digits and a currency symbol. Compared with the reference script without normalising first, this disagrees on every number in the utterance and a word error rate computed that way reports a fault that does not exist. That gap is the point of the pair.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
txt
280 BStaff briefing: reference script
Exactly what was spoken, which is what the synthesiser was given: numbers, currency and times are spelled out as words because that is how they were said. This is ground truth by construction - it existed before the audio did - and it is deliberately NOT the same string as the recogniser transcript paired with it.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
srt
451 BStaff briefing: timed captions
The recogniser's output with timings, 6 cues over the full take. Carries the same normalisation as the plain transcript in this group, so it inherits the same scoring trap, and adds the timing dimension: a cue that starts before the word is spoken is a different defect from a cue with the wrong words in it.
- File
- SRT · Speech Ladders · 6 cues
- Use case
- ASR testingMedia accessibility· Conversion set
wav
39.2 KBSupplier order call: chain telephone
A 5 second excerpt of this group's take, telephone chain. Band-limited, resampled and companded in one pass: the whole telephone path applied at once rather than one effect at a time. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
147.6 KBSupplier order call: clipped
A 5 second excerpt of this group's take, clipping. Samples driven past a fixed threshold and flattened, the damage a hot microphone does and the one no codec can undo. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBSupplier order call: codec g722 16k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as adpcm_g722 at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · adpcm_g722
- Use case
- ASR testingAudio analysis+1· Conversion set
gsm
8.1 KBSupplier order call: codec gsm 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as gsm at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- GSM · Speech Ladders · gsm
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
79.5 KBSupplier order call: codec mp3 128
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
mp3
20 KBSupplier order call: codec mp3 32
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as mp3 at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- MP3 · Speech Ladders · mp3
- Use case
- ASR testingAudio analysis+1· Conversion set
wav
39.2 KBSupplier order call: codec mulaw 8k
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as pcm_mulaw at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- WAV · Speech Ladders · pcm_mulaw
- Use case
- ASR testingAudio analysis+1· Conversion set
opus
15.3 KBSupplier order call: codec opus 24
A 5 second excerpt of this group's take, codec. Encoded and decoded through a real speech codec, so the artefacts are the ones the format actually produces rather than a simulation of them. Delivered as opus at 48000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- OPUS · Speech Ladders · opus
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
138 KBSupplier order call: dropouts
A 5 second excerpt of this group's take, dropouts. Short spans silenced outright, as a packet-switched call drops them, so a transcriber must decide between a pause and a missing word. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
610.5 KBSupplier order call: full spoken take
The complete 22.2 second utterance at 24000 Hz mono FLAC, spoken by the af_sarah voice at 1.0x rate. The five-second excerpt in this group's ladder is cut from this take, so anything needing real duration uses this file and anything comparing damage uses the ladder. Lossless, so it is the archival reference for the whole group.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis· Conversion set
flac
328.6 KBSupplier order call: noise snr0
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
309.1 KBSupplier order call: noise snr10
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
293.4 KBSupplier order call: noise snr20
A 5 second excerpt of this group's take, noise. Broadband noise mixed in at a measured signal-to-noise ratio, which is where a denoiser is either working or not. Delivered as flac at 24000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
147.5 KBSupplier order call: plate
A 5 second excerpt of this group's take, reference. The clean five-second excerpt every other file in this ladder is scored against. Delivered as flac at 24000 Hz mono. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
186.7 KBSupplier order call: rate 16000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 16000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
390.4 KBSupplier order call: rate 44100
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 44100 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
flac
98.1 KBSupplier order call: rate 8000
A 5 second excerpt of this group's take, resample. The same audio at a different sample rate, which is the step most speech pipelines get wrong silently by assuming their model's rate. Delivered as flac at 8000 Hz mono. Scored against plate.flac in the same group, which is the same five seconds undamaged. The parameters in the specification are what the file MEASURED, not what was requested of it.
- File
- FLAC · Speech Ladders · flac
- Use case
- ASR testingAudio analysis+1· Conversion set
txt
332 BSupplier order call: recogniser transcript
What a speech recogniser returned for this take. It normalises: spoken "forty three dollars and eighteen cents" comes back as digits and a currency symbol. Compared with the reference script without normalising first, this disagrees on every number in the utterance and a word error rate computed that way reports a fault that does not exist. That gap is the point of the pair.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
txt
381 BSupplier order call: reference script
Exactly what was spoken, which is what the synthesiser was given: numbers, currency and times are spelled out as words because that is how they were said. This is ground truth by construction - it existed before the audio did - and it is deliberately NOT the same string as the recogniser transcript paired with it.
- File
- TXT · Speech Ladders
- Use case
- ASR testing· Paired fixture
srt
521 BSupplier order call: timed captions
The recogniser's output with timings, 6 cues over the full take. Carries the same normalisation as the plain transcript in this group, so it inherits the same scoring trap, and adds the timing dimension: a cue that starts before the word is spoken is a different defect from a cue with the wrong words in it.
- File
- SRT · Speech Ladders · 6 cues
- Use case
- ASR testingMedia accessibility· Conversion set
Stereo Mono
wav
21.9 KBStereo/Mono Twin 01: Mono (440 Hz)
Mono 440 Hz twin. Pair with the stereo file for channel-count / downmix tests.
- File
- WAV · Stereo Mono · 16000
- Use case
- Audio analysisConversion testing· Paired fixture
wav
43.8 KBStereo/Mono Twin 01: Stereo (440 Hz)
Stereo twin of the 440 Hz tone (L=mono, R=phase-shifted) for stereo↔mono conversion tests.
- File
- WAV · Stereo Mono · 16000
- Use case
- Audio analysisConversion testing· Paired fixture
wav
21.9 KBStereo/Mono Twin 02: Mono (523 Hz)
Mono 523 Hz twin. Pair with the stereo file for channel-count / downmix tests.
- File
- WAV · Stereo Mono · 16000
- Use case
- Audio analysisConversion testing· Paired fixture
wav
43.8 KBStereo/Mono Twin 02: Stereo (523 Hz)
Stereo twin of the 523 Hz tone (L=mono, R=phase-shifted) for stereo↔mono conversion tests.
- File
- WAV · Stereo Mono · 16000
- Use case
- Audio analysisConversion testing· Paired fixture
wav
21.9 KBStereo/Mono Twin 03: Mono (659 Hz)
Mono 659 Hz twin. Pair with the stereo file for channel-count / downmix tests.
- File
- WAV · Stereo Mono · 16000
- Use case
- Audio analysisConversion testing· Paired fixture
wav
43.8 KBStereo/Mono Twin 03: Stereo (659 Hz)
Stereo twin of the 659 Hz tone (L=mono, R=phase-shifted) for stereo↔mono conversion tests.
- File
- WAV · Stereo Mono · 16000
- Use case
- Audio analysisConversion testing· Paired fixture
wav
21.9 KBStereo/Mono Twin 04: Mono (784 Hz)
Mono 784 Hz twin. Pair with the stereo file for channel-count / downmix tests.
- File
- WAV · Stereo Mono · 16000
- Use case
- Audio analysisConversion testing· Paired fixture
wav
43.8 KBStereo/Mono Twin 04: Stereo (784 Hz)
Stereo twin of the 784 Hz tone (L=mono, R=phase-shifted) for stereo↔mono conversion tests.
- File
- WAV · Stereo Mono · 16000
- Use case
- Audio analysisConversion testing· Paired fixture
Tagged
flac
94.2 KBFLAC with Metadata Stripped
The matching FLAC reduced to STREAMINFO and the exact same audio frames, so metadata-removal tests can compare compressed essence byte for byte.
- File
- FLAC · Tagged
- Use case
- Metadata testingConversion testing· Paired fixture
flac
94.6 KBFLAC with Vorbis Comments and Picture
A FLAC whose STREAMINFO and audio frames are accompanied by known Vorbis comments and a 32-pixel PNG front cover.
- File
- FLAC · Tagged
- Use case
- Metadata testingConversion testing· Paired fixture
mp3
67.6 KBID3-tagged MP3 (title / artist / album)
An MP3 with a full set of ID3v2.3 tags (title, artist, album, date, genre, comment) over a synthetic C-major melody. A fixture for testing tag readers, metadata editors, and music-library importers.
- File
- MP3 · Tagged · MP3 (LAME)
- Use case
- Audio analysisConversion testing
mp3
35 KBMP3 with ID3 Metadata Stripped
The same MPEG audio frames with leading and trailing ID3 metadata removed, for checking metadata scrubbing without an audio transcode.
- File
- MP3 · Tagged
- Use case
- Metadata testingConversion testing· Paired fixture
mp3
35.5 KBMP3 with ID3 Tags and Cover Art
An MP3 with deterministic ID3v2.3 text frames and embedded PNG cover art wrapped around unchanged MPEG audio frames.
- File
- MP3 · Tagged
- Use case
- Metadata testingConversion testing· Paired fixture