Whisper Docker Setup

This is a Docker setup with a FastAPI proxy that provides basic authentication for the Whisper ASR (Automatic Speech Recognition) service.

Prerequisites

• Docker and Docker Compose
• Access to the ITKDev Docker commands (idc)

Setup

1. Copy the example environment file and configure it for your needs:

cp .env.example .env

2. Update the API key and other configuration in the .env file:

   • API_KEY: Set a secure API key for authentication
   • ASR_ENGINE: Choose the ASR engine (default: openai_whisper)
   • ASR_MODEL: Set the model to use (default: large-v3)

3. Build and start the containers:

idc build --pull --no-cache
idc up -d

Configuration Options

The following environment variables can be configured in your .env file:

• COMPOSE_PROJECT_NAME: Project name for Docker Compose (default: whisper)
• COMPOSE_DOMAIN: Domain for local development (default: whisper.local.itkdev.dk)
• API_KEY: Authentication key for the API
• WHISPER_ENDPOINT: Internal endpoint for the Whisper service
• ASR_ENGINE: Speech recognition engine to use
• ASR_MODEL: Model to use for speech recognition

API Documentation

The API documentation is available at /docs after starting the services.

Links

• Whisper ASR Webservice Documentation
• FastAPI Documentation
• OpenAI Speech to text API