# Vaibhavs10/insanely-fast-whisper

Repository: https://github.com/Vaibhavs10/insanely-fast-whisper
Canonical: https://ross.abutalabs.com/products/insanely-fast-whisper
Language: Jupyter Notebook
License: Apache-2.0
License Family: permissive
Last push: 2025-10-25T09:50:57+00:00

## Health v2 (maintenance only)
Score: 49/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 48, release rhythm 35, longevity 75
- inputs: {"age_days": 1058, "days_push": 312, "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 13049, forks 960 (observed 2026-08-28T04:11:01.804051+00:00)

## What it is
A CLI tool for transcribing audio files on-device using OpenAI's Whisper models, powered by Hugging Face Transformers, Optimum, and Flash Attention 2. It achieves extremely fast transcription speeds (e.g., 150 minutes of audio in under 2 minutes on an A100 GPU) and supports GPU and Apple MPS devices.

## Use cases
- transcribe audio files to text from the terminal
- convert podcast recordings into transcripts
- generate subtitles from video or audio
- batch transcribe long audio quickly on a GPU
- run Whisper large-v3 locally without cloud APIs
- transcribe audio on a Mac with Apple Silicon MPS

## When to choose
- you need fast, on-device speech-to-text with Whisper models
- you have a CUDA GPU or Apple Silicon and want to avoid cloud transcription costs
- you want a simple pipx-installable CLI rather than writing Python code
- you need to transcribe long recordings in minutes instead of hours

## When to avoid
- you have no GPU or Apple Silicon and need fast transcription
- you need a full-featured transcription service with speaker diarization and editing
- you want a GUI application rather than a command-line tool
- you need real-time streaming transcription

## Facets
- artifact type: cli-tool
- maturity: active
- function: speech-recognition, cli, machine-learning, llm-inference
- domain: speech-processing, artificial-intelligence
- platform: cli, python, windows
- tags: whisper, transcription, audio-to-text, transformers, flash-attention, hugging-face, natural-language-processing, command-line, macos, linux, gpu

## Member repositories
- Vaibhavs10/insanely-fast-whisper (main) score 49

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:01.804051+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 2026-08-29T17:13:34.494737+00:00, confidence not recorded.
  - readme: https://github.com/Vaibhavs10/insanely-fast-whisper (fetched 2026-08-28T04:11:01.804051+00:00, sha 303bb7eb0384)
  - registry_pypi: https://pypi.org/pypi/insanely-fast-whisper/json (fetched 2026-08-29T08:09:12.712870+00:00, sha 75ca0c9a580f)
- Data as of 2026-08-30T08:39:29.467469+00:00.