# m-bain/whisperX

WhisperX:  Automatic Speech Recognition with Word-level Timestamps (& Diarization)

Repository: https://github.com/m-bain/whisperX
Canonical: https://ross.abutalabs.com/products/whisperx
Language: Python
License: BSD-2-Clause
License Family: permissive
Topics: asr, speech, speech-recognition, speech-to-text, whisper
Last push: 2026-07-13T08:30:07+00:00

## Health v2 (maintenance only)
Score: 91/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 92, release rhythm 85, longevity 97
- inputs: {"age_days": 1364, "days_push": 51, "days_rel": 100, "gap_med": 1, "n_releases_24m": 36}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 23765, forks 2387 (observed 2026-08-28T04:11:35.338006+00:00)

## What it is
WhisperX is a Python library providing fast automatic speech recognition with word-level timestamps and speaker diarization. It combines Whisper-based batched inference (up to 70x realtime), wav2vec2 forced alignment for accurate word timings, VAD preprocessing, and pyannote-audio speaker labeling.

## Use cases
- transcribe audio files with word-level timestamps
- add speaker labels to meeting or interview transcripts
- run fast batched speech-to-text on GPU
- reduce Whisper hallucinations with voice activity detection
- align existing transcripts to audio at phoneme level
- diarize multi-speaker recordings

## When to choose
- you need per-word timing accuracy beyond Whisper's utterance-level timestamps
- you want fast GPU transcription with batching and a memory-efficient backend
- you need speaker diarization integrated into the ASR pipeline

## When to avoid
- you need a hosted transcription API with real speaker names from meeting platforms
- you only need rough utterance-level transcripts and want a simpler setup
- you have no GPU and need maximum throughput

## Facets
- artifact type: library
- maturity: active
- function: speech-recognition, machine-learning, audio-processing
- domain: speech-processing, machine-learning
- platform: python, cli
- tags: asr, whisper, word-level-timestamps, speaker-diarization, forced-alignment, transcription, faster-whisper, wav2vec2, pyannote, audio, gpu

## Member repositories
- m-bain/whisperX (main) score 91

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:35.338006+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 2026-08-29T16:56:21.055951+00:00, confidence not recorded.
  - readme: https://github.com/m-bain/whisperX (fetched 2026-08-28T04:11:35.338006+00:00, sha ec77381252da)
  - registry_pypi: https://pypi.org/pypi/whisperx/json (fetched 2026-08-29T07:53:56.868266+00:00, sha 20531f2c2abd)
- Data as of 2026-08-30T08:39:29.467469+00:00.