Skip to content
Get packs

Whisper

Open-source speech-to-text model that transcribes and translates audio in many languages.

GitHub Download v20250625
Stars
112k
Forks
13.5k
Licence
MIT
Language
Python
Last update
2 months ago
Type
AI model / agent

What it is

Whisper is a speech recognition model from OpenAI that turns spoken audio into written text. It was trained on a large set of varied audio and can also translate non-English speech into English and identify which language is being spoken. You can use it from a command line or from Python code.

  • Transcribe audio files such as MP3, WAV, or FLAC from the command line
  • Translate non-English speech into English using a multilingual model
  • Detect the spoken language of an audio clip from Python
  • Generate text from recordings inside your own Python script

Great for

  • Developers who want to add transcription of audio files to a Python project
  • People who need to transcribe recordings such as interviews or lectures locally
  • Projects needing multilingual transcription with one model
  • Users who want English-only transcription with the smaller .en models

Not ideal for

  • Users without a suitable computer, since larger models need several GB of video memory (VRAM), per the README's table
  • Anyone needing translation with the turbo model, which the README says is not trained for translation
  • People wanting a point-and-click app, since the README describes command-line and Python usage only
Step-by-step guide: how to use WhisperInstall, first run, examples, tips and FAQs
VibeVoice vs WhisperWhisper vs whisper.cppfaster-whisper vs WhisperFunASR vs Whisper

How to get it

  1. 1

    Install with a package manager

    You'll need: Python 3.8-3.11 (per README compatibility note) · ffmpeg · Rust (only if tiktoken has no pre-built wheel for your platform)

    Install the latest release

    pip install -U openai-whisper

    Install ffmpeg on Ubuntu or Debian

    sudo apt update && sudo apt install ffmpeg

    Install ffmpeg on MacOS with Homebrew

    brew install ffmpeg
  2. 2

    Run it

    • Install Python, ffmpeg, and the openai-whisper package
    • Run whisper audio.mp3 --model turbo on an audio file
    • Check the output text, or run whisper --help to see all options
  3. 3

    Get the most out of it

    • Use a multilingual model such as medium or large with --task translate to translate non-English speech into English
    • Specify --language when you know the spoken language to avoid automatic detection
    • Use the .en models for English-only audio, especially tiny.en and base.en, per the README
    • Call model.transcribe() from Python to process files in your own code
    • Browse the Show and tell discussions for third-party integrations and ports

Downloads

Source code (.zip)main branch · straight from GitHubRelease v20250625Release notes and source

Strengths

  • Handles many languages and can translate speech to English in one model
  • Offers several model sizes so you can trade speed against accuracy
  • Released under the MIT License, which permits broad reuse

Trade-offs

  • Larger models need roughly 5 to 10 GB of VRAM, so modest hardware may struggle
  • Accuracy varies widely by language, according to the README
  • Requires ffmpeg and sometimes Rust for installation

Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.