@voxextractlabs/vox-whisper

Project Summary and Key Features

• Fast Transcription: A lightweight wrapper around the Faster-Whisper Docker CLI.
• Cross-Platform: Designed to work with Node.js and Bun.
• Configurable: Supports CPU and GPU inference along with various runtime options.
• Modern Tooling: Integrates Biome, conventional commits, lefthook, and more.

Installation

Install the package via bun or your preferred package manager:

bun add @voxextractlabs/vox-whisper
yarn add @voxextractlabs/vox-whisper
npm install @voxextractlabs/vox-whisper

Usage Example

import { whisperTranscribe } from '@voxextractlabs/vox-whisper';

const result = await whisperTranscribe({
    input: 'path/to/audio.wav',
    output: 'tmp/output/result.txt',
    device: 'cpu',
    dockerImage: 'voxextractlabs/vox-whisper:cpu-v1.0.0'
});

console.log('Transcription result:', result);

Configuration Details

The function accepts a WhisperConfig object with the following fields:

| Property | Type | Description | |------------------|--------------------|----------------------------------------------| | input | string | Path to the input audio file | | output | string | Path where the transcription will be written | | device | 'cpu' \| 'cuda' | Inference device | | dockerImage | string | Docker image to use | | model | string? | Model size: tiny, base, small, etc. | | language | string? | Language code (e.g. en) | | output_format | string? | Format: txt, json, srt | | timestamps | boolean? | Include word-level timestamps (JSON only) | | threads | number? | Number of CPU threads | | verbose | boolean? | Enable verbose/debug logging |

Contribution Guidelines

Contributions are welcome! Please follow the Conventional Commits specification and run pre-commit hooks using Lefthook. Ensure that your code passes all lint checks, tests, and includes docs where appropriate.

License and Acknowledgments

Licensed under the MIT License.
Special thanks to the contributors and the open-source community for their invaluable support.