nyrahealth/CrisperWhisper

Controllable Transcription. Verbatim ( every, filler, pause, stutter, vocal sound) , or intended ( what the speaker meant to say, optimized for readability) with word-level timestamps.

1,390

stars

47

commits

Python

primary language

Aug 23, 2026

updated

nyra-labs.com
asr
audio
detection
filler
full-duplex-audio
labeling
recognition
speech
speech-processing
speech-recognition
stutter-detection
stuttering
timestamps
transcription
turn-taking
verbatim
whisper

README

CrisperWhisper 2.0

PyPI

The most accurate verbatim speech recognition you can run in production: controllable, multilingual, and timed to the word.

Release post · Paper · Full documentation · Models · Benchmark

Try it now

Most speech-to-text systems never actually decide whether to write down what was said or what was meant. They inherit that choice from their training data and apply it inconsistently. CrisperWhisper 2.0 makes it an explicit, controllable choice. One recording, two transcripts:

Verbatim, exactly what was said, in one consistent format: [um] so we we need to, to reschedule the th- thursday meeting to [uh] march third at nine thirty [laughter]

Intended, the clean version the speaker meant, with numbers, dates, and emails formatted the way you'd write them: So we need to reschedule the Thursday meeting to March 3 at 9:30.

On top of that:

  • Word-level timings. Around 30 ms mean boundary error on read speech and 41 ms on conversational speech, the most precise word timing of any system we benchmarked, on both.
  • Verbatimize. Upgrade transcripts you already have: given audio plus a trusted clean transcript, the model reproduces your content word-for-word and inserts only the disfluencies and vocal events actually present in the audio (rare-word recall jumps from 6.8% to 96.1% vs. re-transcribing). This turns the world's abundant clean corpora into verbatim ones, ready for TTS data, clinical speech analysis, and dataset construction.
  • Multilingual. Verbatim and intended modes work across most languages Whisper supports. CrisperWhisper 2.0 tops the Nyra Verbatim Speech Benchmark leaderboard for disfluency F1 across ten languages, ahead of every closed-source alternative we tested.
  • Seamless longform. Audio of any length, transcribed without the usual chunk-boundary artifacts: each window continues from the words already transcribed (conditional continuation), so there are no duplicated or dropped words at the seams and no fragile timestamp-token bookkeeping.
  • Production inference. A CTranslate2 runtime with speculative decoding and built-in mitigation of Whisper's looping-hallucination failure mode.

Performance

The Nyra Verbatim Speech Benchmark scores fillers, repetitions, cut-offs, and vocal sounds as separate, typed metrics. Its headline number is disfluency F1: how reliably a system writes down the disfluencies that were actually spoken, without inventing ones that weren't. Averaged over ten languages:

#SystemDisfluency F1
1CrisperWhisper 2.0 Pro93.5
2CrisperWhisper 2.087.8
3ElevenLabs Scribe v279.2
4Microsoft MAI-Transcribe-1.577.5
5CrisperWhisper 1.0*64.8
6Inworld STT59.5
7xAI Grok Speech-to-Text42.8
8Deepgram Nova-337.8
9Fish Audio ASR35.0
10AssemblyAI Universal-3 Pro30.5

* CrisperWhisper 1.0 is English/German-only; its average covers those two languages. English and German use human-labeled evaluation sets; the other eight languages use synthetic verbatim sets. Per-language breakdowns and how the metric is computed are in the benchmark post.

Word-timing accuracy

Mean absolute word-boundary error on read speech (TIMIT), lower is better:

#SystemBoundary error
1CrisperWhisper 2.029.6 ms
2xAI Grok Speech-to-Text37.1 ms
3CTC-seg49.3 ms
4ElevenLabs Scribe v251.3 ms
5NeMo-FA60.0 ms
6Deepgram Nova-363.3 ms
7WhisperX64.8 ms
8Cartesia Ink-Whisper69.4 ms
9Canary85.5 ms

Scored on exactly the words each system gets right. How the timings are extracted from supervised cross-attention, plus results on conversational speech, are in the aligner post.

Install

# NVIDIA GPU (Linux): fastest, includes speculative decoding.
# An NVIDIA driver is all you need; CUDA libraries arrive via pip.
pip install "crisperwhisper[ct2]"

# Pure PyTorch: runs anywhere torch does (macOS, Windows, CPU)
pip install "crisperwhisper[transformers]"

Quickstart

from crisperwhisper import CrisperWhisperModel

model = CrisperWhisperModel()          # nyralabs/CrisperWhisper2.0_large
# or pick a size: CrisperWhisperModel("turbo")  # turbo / medium / small

# Verbatim transcription (default): every filler, repetition, stutter,
# false start, and vocal event
result = model.transcribe("meeting.wav", language="en")
print(result.text)

# Intended: the clean, readable version
clean = model.transcribe("meeting.wav", language="en", mode="intended")

# Word-level timestamps
result = model.transcribe("meeting.wav", language="en", word_timestamps=True)
for w in result.words:
    print(f"{w.start:6.2f}-{w.end:6.2f}  {w.word}")

# Verbatimize: upgrade an existing clean transcript with the
# disfluencies that are actually in the audio
result = model.verbatimize("clip.wav", "I think we should ship it Friday.")

Audio longer than 30 seconds is handled automatically (see longform below). The first load of a model downloads it from HuggingFace and, on the ct2 backend, converts it once into a local cache. That one-time conversion needs torch and transformers, which the lean [ct2] extra does not install -- add them once with:

pip install "crisperwhisper[ct2,convert]"   # or add [convert] later

A CPU-only torch is enough for the conversion, and the dependencies can be uninstalled once the converted model is cached. Loading a directory that already contains a converted CT2 model (model.bin) never needs them.

Models

ShorthandHuggingFace IDNotes
"large" (default)nyralabs/CrisperWhisper2.0_largeBest open quality
"turbo"nyralabs/CrisperWhisper2.0_turboFastest, with some quality degradation; recommended as the speculative draft
"medium"nyralabs/CrisperWhisper2.0_mediumNear-large quality; best tradeoff between size and quality
"small"nyralabs/CrisperWhisper2.0_smallSmallest
"large_pro" / "turbo_pro" / "medium_pro" / "small_pro"nyralabs/CrisperWhisper2.0_<size>_proPro: our best models, with improved performance, hotword boosting, trained on additional proprietary data

The standard models are released under a non-commercial research license and are available for commercial licensing. The Pro models are available under commercial license only. For both, get in touch.

Faster inference: speculative decoding (ct2)

A small draft model proposes tokens and the main model verifies them. Same output, 1.3 to 1.4x faster:

model = CrisperWhisperModel("large", draft_model="turbo")
result = model.transcribe("meeting.wav", language="en",
                          speculative_decoding=True)

What else is in the box

Everything below works out of the box and is covered in depth in DOCS.md:

OptionWhat it does
mode="verbatim" / "intended"Choose what-was-said vs. what-was-meant per call
word_timestamps=TruePer-word start/end times from supervised cross-attention alignment
hotwords=[...]Bias recognition toward names and rare terms (Pro models only)
model.transcribe_dual(...)Verbatim and intended in one pass (ct2)
model.verbatimize(audio, transcript)Insert real disfluencies into a trusted clean transcript
model.forced_align(audio, text)Timings for a transcript you already have
LongformAudio >30s transcribed seamlessly via conditional continuation, with no chunk-boundary duplicates, drops, or stitching
Hallucination mitigationOn by default: detects and suppresses Whisper's looping-repetition failure mode during decoding
compute_type="float16" / "int8_float16"Quantization

How it works

Each mechanism has a deep-dive post:

Documentation

DOCS.md covers the full API: backends and their trade-offs, every transcribe() option, dual-mode transcription, forced alignment, longform strategies, speculative-K tuning, hallucination-repair thresholds, quantization, model conversion, and the result object.

License

The inference code in this repository is MIT-licensed (see LICENSE): use it freely, commercially or otherwise. crisperwhisper/features.py is vendored from faster-whisper (MIT, SYSTRAN).

The model weights are not MIT. They are released under the Nyra Health Non-Commercial Research License: free for research and other non-commercial use; any commercial use requires a commercial license. The Pro models are available under commercial license only. For commercial licensing of either, contact Nyra.

Contributors

LaurinmyReha

40 commits

tsmdt

2 commits

hlevring

1 commits

nyrahealth/CrisperWhisper

Controllable Transcription. Verbatim ( every, filler, pause, stutter, vocal sound) , or intended ( what the speaker meant to say, optimized for readability) with word-level timestamps.

1,390

stars

47

commits

Python

primary language

Aug 23, 2026

updated

nyra-labs.com
asr
audio
detection
filler
full-duplex-audio
labeling
recognition
speech
speech-processing
speech-recognition
stutter-detection
stuttering
timestamps
transcription
turn-taking
verbatim
whisper

README

CrisperWhisper 2.0

PyPI

The most accurate verbatim speech recognition you can run in production: controllable, multilingual, and timed to the word.

Release post · Paper · Full documentation · Models · Benchmark

Try it now

Most speech-to-text systems never actually decide whether to write down what was said or what was meant. They inherit that choice from their training data and apply it inconsistently. CrisperWhisper 2.0 makes it an explicit, controllable choice. One recording, two transcripts:

Verbatim, exactly what was said, in one consistent format: [um] so we we need to, to reschedule the th- thursday meeting to [uh] march third at nine thirty [laughter]

Intended, the clean version the speaker meant, with numbers, dates, and emails formatted the way you'd write them: So we need to reschedule the Thursday meeting to March 3 at 9:30.

On top of that:

  • Word-level timings. Around 30 ms mean boundary error on read speech and 41 ms on conversational speech, the most precise word timing of any system we benchmarked, on both.
  • Verbatimize. Upgrade transcripts you already have: given audio plus a trusted clean transcript, the model reproduces your content word-for-word and inserts only the disfluencies and vocal events actually present in the audio (rare-word recall jumps from 6.8% to 96.1% vs. re-transcribing). This turns the world's abundant clean corpora into verbatim ones, ready for TTS data, clinical speech analysis, and dataset construction.
  • Multilingual. Verbatim and intended modes work across most languages Whisper supports. CrisperWhisper 2.0 tops the Nyra Verbatim Speech Benchmark leaderboard for disfluency F1 across ten languages, ahead of every closed-source alternative we tested.
  • Seamless longform. Audio of any length, transcribed without the usual chunk-boundary artifacts: each window continues from the words already transcribed (conditional continuation), so there are no duplicated or dropped words at the seams and no fragile timestamp-token bookkeeping.
  • Production inference. A CTranslate2 runtime with speculative decoding and built-in mitigation of Whisper's looping-hallucination failure mode.

Performance

The Nyra Verbatim Speech Benchmark scores fillers, repetitions, cut-offs, and vocal sounds as separate, typed metrics. Its headline number is disfluency F1: how reliably a system writes down the disfluencies that were actually spoken, without inventing ones that weren't. Averaged over ten languages:

#SystemDisfluency F1
1CrisperWhisper 2.0 Pro93.5
2CrisperWhisper 2.087.8
3ElevenLabs Scribe v279.2
4Microsoft MAI-Transcribe-1.577.5
5CrisperWhisper 1.0*64.8
6Inworld STT59.5
7xAI Grok Speech-to-Text42.8
8Deepgram Nova-337.8
9Fish Audio ASR35.0
10AssemblyAI Universal-3 Pro30.5

* CrisperWhisper 1.0 is English/German-only; its average covers those two languages. English and German use human-labeled evaluation sets; the other eight languages use synthetic verbatim sets. Per-language breakdowns and how the metric is computed are in the benchmark post.

Word-timing accuracy

Mean absolute word-boundary error on read speech (TIMIT), lower is better:

#SystemBoundary error
1CrisperWhisper 2.029.6 ms
2xAI Grok Speech-to-Text37.1 ms
3CTC-seg49.3 ms
4ElevenLabs Scribe v251.3 ms
5NeMo-FA60.0 ms
6Deepgram Nova-363.3 ms
7WhisperX64.8 ms
8Cartesia Ink-Whisper69.4 ms
9Canary85.5 ms

Scored on exactly the words each system gets right. How the timings are extracted from supervised cross-attention, plus results on conversational speech, are in the aligner post.

Install

# NVIDIA GPU (Linux): fastest, includes speculative decoding.
# An NVIDIA driver is all you need; CUDA libraries arrive via pip.
pip install "crisperwhisper[ct2]"

# Pure PyTorch: runs anywhere torch does (macOS, Windows, CPU)
pip install "crisperwhisper[transformers]"

Quickstart

from crisperwhisper import CrisperWhisperModel

model = CrisperWhisperModel()          # nyralabs/CrisperWhisper2.0_large
# or pick a size: CrisperWhisperModel("turbo")  # turbo / medium / small

# Verbatim transcription (default): every filler, repetition, stutter,
# false start, and vocal event
result = model.transcribe("meeting.wav", language="en")
print(result.text)

# Intended: the clean, readable version
clean = model.transcribe("meeting.wav", language="en", mode="intended")

# Word-level timestamps
result = model.transcribe("meeting.wav", language="en", word_timestamps=True)
for w in result.words:
    print(f"{w.start:6.2f}-{w.end:6.2f}  {w.word}")

# Verbatimize: upgrade an existing clean transcript with the
# disfluencies that are actually in the audio
result = model.verbatimize("clip.wav", "I think we should ship it Friday.")

Audio longer than 30 seconds is handled automatically (see longform below). The first load of a model downloads it from HuggingFace and, on the ct2 backend, converts it once into a local cache. That one-time conversion needs torch and transformers, which the lean [ct2] extra does not install -- add them once with:

pip install "crisperwhisper[ct2,convert]"   # or add [convert] later

A CPU-only torch is enough for the conversion, and the dependencies can be uninstalled once the converted model is cached. Loading a directory that already contains a converted CT2 model (model.bin) never needs them.

Models

ShorthandHuggingFace IDNotes
"large" (default)nyralabs/CrisperWhisper2.0_largeBest open quality
"turbo"nyralabs/CrisperWhisper2.0_turboFastest, with some quality degradation; recommended as the speculative draft
"medium"nyralabs/CrisperWhisper2.0_mediumNear-large quality; best tradeoff between size and quality
"small"nyralabs/CrisperWhisper2.0_smallSmallest
"large_pro" / "turbo_pro" / "medium_pro" / "small_pro"nyralabs/CrisperWhisper2.0_<size>_proPro: our best models, with improved performance, hotword boosting, trained on additional proprietary data

The standard models are released under a non-commercial research license and are available for commercial licensing. The Pro models are available under commercial license only. For both, get in touch.

Faster inference: speculative decoding (ct2)

A small draft model proposes tokens and the main model verifies them. Same output, 1.3 to 1.4x faster:

model = CrisperWhisperModel("large", draft_model="turbo")
result = model.transcribe("meeting.wav", language="en",
                          speculative_decoding=True)

What else is in the box

Everything below works out of the box and is covered in depth in DOCS.md:

OptionWhat it does
mode="verbatim" / "intended"Choose what-was-said vs. what-was-meant per call
word_timestamps=TruePer-word start/end times from supervised cross-attention alignment
hotwords=[...]Bias recognition toward names and rare terms (Pro models only)
model.transcribe_dual(...)Verbatim and intended in one pass (ct2)
model.verbatimize(audio, transcript)Insert real disfluencies into a trusted clean transcript
model.forced_align(audio, text)Timings for a transcript you already have
LongformAudio >30s transcribed seamlessly via conditional continuation, with no chunk-boundary duplicates, drops, or stitching
Hallucination mitigationOn by default: detects and suppresses Whisper's looping-repetition failure mode during decoding
compute_type="float16" / "int8_float16"Quantization

How it works

Each mechanism has a deep-dive post:

Documentation

DOCS.md covers the full API: backends and their trade-offs, every transcribe() option, dual-mode transcription, forced alignment, longform strategies, speculative-K tuning, hallucination-repair thresholds, quantization, model conversion, and the result object.

License

The inference code in this repository is MIT-licensed (see LICENSE): use it freely, commercially or otherwise. crisperwhisper/features.py is vendored from faster-whisper (MIT, SYSTRAN).

The model weights are not MIT. They are released under the Nyra Health Non-Commercial Research License: free for research and other non-commercial use; any commercial use requires a commercial license. The Pro models are available under commercial license only. For commercial licensing of either, contact Nyra.

Contributors

LaurinmyReha

40 commits

tsmdt

2 commits

hlevring

1 commits

Languages

Python

100.0%