# 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.

Repository: https://github.com/nyrahealth/CrisperWhisper
Canonical: https://ross.abutalabs.com/products/crisperwhisper
Homepage: https://nyra-labs.com
Language: Python
License: NOASSERTION
License Family: other
Topics: asr, audio, detection, filler, recognition, speech, speech-recognition, timestamps, transcription, verbatim, whisper, speech-processing, stutter-detection, stuttering, full-duplex-audio, labeling, turn-taking
Last push: 2026-08-23T13:18:11+00:00

## Health v2 (maintenance only)
Score: 89/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 94, longevity 59
- inputs: {"age_days": 831, "days_push": 10, "days_rel": 41, "gap_med": 2, "n_releases_24m": 2}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1349, forks 86 (observed 2026-08-28T04:04:27.842034+00:00)

## What it is
CrisperWhisper 2.0 is a controllable speech recognition model and Python library that transcribes audio either verbatim (including fillers, repetitions, stutters, and vocal sounds) or in a cleaned 'intended' form, with precise word-level timestamps. It builds on Whisper with a fast CTranslate2 runtime, multilingual support, and seamless longform transcription.

## Use cases
- transcribe audio verbatim with fillers and stutters included
- generate clean readable transcripts from speech
- get word-level timestamps for subtitles or alignment
- convert clean transcript corpora into verbatim datasets for TTS training
- analyze disfluencies in clinical speech or stuttering research
- transcribe long recordings without chunk-boundary artifacts

## When to choose
- you need verbatim transcription that captures disfluencies, fillers, and vocal events
- precise word-level timing matters, e.g. for subtitles, labeling, or speech analysis
- you want a controllable choice between verbatim and cleaned output in one model
- you need multilingual speech recognition with disfluency detection

## When to avoid
- you only need lightweight real-time dictation with minimal compute
- you require a permissively licensed model for commercial redistribution without checking the custom license
- you need speaker diarization or full-duplex conversation analytics out of the box

## Facets
- artifact type: library
- maturity: active
- function: speech-recognition, nlp, machine-learning, llm-inference
- domain: speech-processing, machine-learning, healthcare
- platform: python, cross-platform
- tags: asr, whisper, verbatim-transcription, word-level-timestamps, disfluency-detection, cttranslate2, filler-detection, stutter-detection, multilingual, natural-language-processing, gpu

## Member repositories
- nyrahealth/CrisperWhisper (main) score 89

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:27.842034+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:42:22.384798+00:00, confidence not recorded.
  - readme: https://github.com/nyrahealth/CrisperWhisper (fetched 2026-08-28T04:04:27.842034+00:00, sha b7137e356cc1)
  - homepage: https://nyra-labs.com (fetched 2026-08-29T12:01:20.605906+00:00, sha 01df6d65c9e4)
  - site_page: https://nyra-labs.com/about (fetched 2026-08-29T12:01:20.615024+00:00, sha ea49edaf033e)
  - registry_pypi: https://pypi.org/pypi/crisperwhisper/json (fetched 2026-08-29T12:01:20.616760+00:00, sha b21768fc4993)
- Data as of 2026-08-30T08:39:29.467469+00:00.