Bazarr LLM Translate

Advanced subtitle translator with LLM support and web API. Converts SRT subtitle files to styled ASS format with bilingual or monolingual translations using OpenAI, Google Gemini, or DeepSeek APIs.

Features

• Web API: FastAPI-based REST API for subtitle translation
• CLI Tool: Command-line interface for batch processing
• Multiple AI Providers: OpenAI, Google Gemini, DeepSeek support
• Bilingual Mode: Displays original text on top and translated text below
• Monolingual Mode: Replaces original text with translation
• Smart Translation: Full text or selective difficulty translation modes
• Resumable: Automatically saves progress and can resume if interrupted
• Batch Processing: Processes subtitles in batches for efficient API usage
• Containerized: Docker support with multi-platform builds

Quick Start

Using Docker (Recommended)

1. Pull the image:

   docker pull ghcr.io/yanp/llm-subtitle-translator:latest
   

2. Run the container:

   docker run -d \
     -p 8080:8080 \
     -e DEEPSEEK_API_KEY=your_api_key_here \
     ghcr.io/yanp/llm-subtitle-translator:latest
   

3. Access the API:

   • Web interface: http://localhost:8080/docs
   • Health check: http://localhost:8080/health

Using Docker Compose

1. Clone the repository:

   git clone https://github.com/yanp/llm-subtitle-translator.git
   cd llm-subtitle-translator
   

2. Set up environment variables:

   cp .env.example .env
   # Edit .env with your API keys
   

3. Start the service:

   docker-compose up -d
   

Local Development

1. Install uv:

   curl -LsSf https://astral.sh/uv/install.sh | sh
   

2. Install dependencies:

   uv venv
   uv pip install -e .
   

3. Set environment variables:

   export DEEPSEEK_API_KEY=your_api_key_here
   

4. Run the API:

   uv run fastapi dev src/llm_subtitle_translator/app.py
   

5. Or use the CLI:

   uv run translate-subtitles input.srt --translation-mode bilingual
   

API Usage

Translate Subtitle File

POST /translate

Supports two modes: file upload or file path processing.

Mode 1: File Upload (Web Interface)

Upload an SRT file and get back a translated ASS file.

Parameters:

• file: SRT subtitle file to upload
• provider: AI provider (openai, gemini, deepseek) - default: deepseek
• model: Specific model name (optional)
• translation_mode: bilingual or monolingual - default: bilingual
• prompt_template: full_text or selective_difficulty - default: selective_difficulty
• batch_size: Number of lines per API call - default: 50

Example:

curl -X POST "http://localhost:8080/translate" \
  -F "file=@subtitle.srt" \
  -F "provider=deepseek" \
  -F "translation_mode=bilingual" \
  -F "prompt_template=selective_difficulty" \
  -o translated_subtitle.ass

Mode 2: File Path (Bazarr Hook)

Process files on the server filesystem (ideal for Bazarr hooks).

Parameters:

• input_path: Path to SRT subtitle file on server
• output_path: Output path for translated file (optional, defaults to same directory with .en-zh.ass extension)
• provider: AI provider (openai, gemini, deepseek) - default: deepseek
• model: Specific model name (optional)
• translation_mode: bilingual or monolingual - default: bilingual
• prompt_template: full_text or selective_difficulty - default: selective_difficulty
• batch_size: Number of lines per API call - default: 50

Example:

curl -X POST "http://localhost:8080/translate" \
  -F "input_path=/media/subtitles/movie.srt" \
  -F "output_path=/media/subtitles/movie.en-zh.ass" \
  -F "provider=deepseek" \
  -F "translation_mode=bilingual"

Get Available Providers

GET /providers

Returns available AI providers and their configuration status.

CLI Usage

uv run translate-subtitles input.srt [options]

Options:

• -o, --output: Output file path
• --translation-mode: bilingual or monolingual
• --prompt-template: full_text or selective_difficulty
• -p, --provider: AI provider (openai, gemini, deepseek)
• -m, --model: Specific model name
• --batch-size: Batch size for API calls

Example:

uv run translate-subtitles movie.srt \
  --translation-mode bilingual \
  --prompt-template selective_difficulty \
  --provider deepseek \
  --batch-size 50

Configuration

Environment Variables

Variable	Description	Required
OPENAI_API_KEY	OpenAI API key	For OpenAI provider
GEMINI_API_KEY	Google Gemini API key	For Gemini provider
DEEPSEEK_API_KEY	DeepSeek API key	For DeepSeek provider

Translation Modes

• Bilingual: Shows original text on top, translation below
• Monolingual: Replaces original text with translation

Prompt Templates

• Full Text: Translates every subtitle line
• Selective Difficulty: Only translates complex phrases, slang, or cultural references

Development

Setup

# Install uv
curl -LsSf https://astral.sh/uv/install.sh | sh

# Install dependencies
uv venv
uv pip install -e ".[dev]"

# Activate virtual environment
source .venv/bin/activate

Code Quality

# Format code
black .

# Lint code
ruff check .

# Type checking
mypy .

Testing

# Run tests
pytest

# Run with coverage
pytest --cov=.

Docker

Build locally

docker build -t llm-subtitle-translator .

Multi-platform build

docker buildx build --platform linux/amd64,linux/arm64 -t llm-subtitle-translator .

License

MIT License - see LICENSE file for details.

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

Support

• Open an issue on GitHub
• Check the API documentation at /docs when running the server
• Review the example files in the repository