Whisper Models

Simple package to download and/or use whisper models in your project, wether for transcription, translation, or any other purpose.

Model	Disk	RAM
tiny	75 MB	~390 MB
tiny.en	75 MB	~390 MB
base	142 MB	~500 MB
base.en	142 MB	~500 MB
small	466 MB	~1.0 GB
small.en	466 MB	~1.0 GB
medium	1.5 GB	~2.6 GB
medium.en	1.5 GB	~2.6 GB
large-v1	2.9 GB	~4.7 GB
large-v2	2.9 GB	~4.7 GB
large-v3	2.9 GB	~4.7 GB

Usage

Install the package using your package manager of choice:

npm install whisper-models
yarn add whisper-models
pnpm add whisper-models

and also add the following line to the scripts object of the package.json depending on the package manager you are using and the model you want to download:

{
  "scripts": {
    "postinstall": "pnpm whisper-models -m small"
  }
}

Transcription

// import whisper from 'whisper-models';
const Whisper = require('whisper-models');

(async () => {
  const whisper = new Whisper('tiny');
  await whisper.run();

  const transcription = await whisper.sendData('path/to/audio/file.wav');
  console.log(transcription);

  // or if you already know the spoken language

  const transcription = await whisper.sendData('path/to/audio/file.wav', { spokenLanguage: 'en' });
  console.log(transcription);
})();

Translation

// import whisper from 'whisper-models';
const Whisper = require('whisper-models');

(async () => {
  const whisper = new Whisper('tiny');
  await whisper.run();

  const translation = await whisper.sendData('path/to/audio/file.wav', { task: 'translate' });
  console.log(translation);
})();

Options

• task: The task to perform. Default is transcribe.
• spokenLanguage: The language spoken in the audio file. Default is en.
• beamSize: The beam size. Default is 5.
• temperature: The sampling temperature (between 0 and 1). Default is 0.
• patience: The patience for early stopping.
• maxSegmentLength: The maximum segment length. Default is 0.
• compressionRatioThreshold: The compression ratio threshold.
• cuda: The Nvidia CUDA device to use. Default is false.