# sanchit-gandhi/whisper-jax

JAX implementation of OpenAI's Whisper model for up to 70x speed-up on TPU.

Repository: https://github.com/sanchit-gandhi/whisper-jax
Canonical: https://ross.abutalabs.com/products/whisper-jax
Language: Jupyter Notebook
License: Apache-2.0
License Family: permissive
Topics: deep-learning, jax, speech-recognition, speech-to-text, whisper
Last push: 2024-04-03T12:12:52+00:00

## Health v2 (maintenance only)
Score: 30/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 91
- inputs: {"age_days": 1280, "days_push": 882, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4682, forks 411 (observed 2026-08-28T04:08:56.897212+00:00)

## What it is
An optimized JAX implementation of OpenAI's Whisper speech recognition model, built on Hugging Face Transformers, offering up to 70x faster transcription than the PyTorch version. It runs on CPU, GPU, and TPU with data parallelism via pmap and can be used as a pipeline or inference endpoint.

## Use cases
- transcribe long audio files fast
- batch transcribe hours of audio on a TPU
- run whisper large v2 with jax
- deploy whisper as an inference endpoint
- speech to text for podcasts or meetings
- benchmark whisper implementations

## When to choose
- you need maximum Whisper throughput on GPU/TPU
- you're transcribing large volumes of audio in batch
- you already use the JAX/Flax ecosystem

## When to avoid
- you need a simple CPU-only transcription tool
- you require PyTorch ecosystem integration or fine-tuning tooling
- you need actively maintained features - the repo is in maintenance mode

## Facets
- artifact type: library
- maturity: maintenance
- function: speech-recognition, machine-learning, llm-inference
- domain: speech-processing, deep-learning, machine-learning
- platform: python, cloud
- tags: jax, whisper, tpu, transcription, asr, hugging-face, gpu

## Member repositories
- sanchit-gandhi/whisper-jax (main) score 30

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:56.897212+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 2026-08-29T18:19:19.400635+00:00, confidence not recorded.
  - readme: https://github.com/sanchit-gandhi/whisper-jax (fetched 2026-08-28T04:08:56.897212+00:00, sha a483df9ad770)
- Data as of 2026-08-30T08:39:29.467469+00:00.