ai-subtitle-translator 0.5.0

Advanced subtitle translator with LLM support for converting SRT to bilingual ASS files using OpenAI, Gemini, or DeepSeek APIs

AI Subtitle Translator

Advanced subtitle translator with AI 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

• Bazarr Integration: Direct post-processing hook for automatic subtitle translation
• 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
• Monolingual Extraction: Convert existing bilingual files to monolingual without re-translating
• 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

Quick Start - Bazarr Integration

1. Docker Compose Setup

1. Clone the repository:

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

2. Set up environment variables:

   cp .env.example .env
   # Edit .env and add: DEEPSEEK_API_KEY=your_actual_api_key_here
   

3. Start the service:

   docker-compose up -d
   

2. Bazarr Hook Configuration

In Bazarr Settings → Subtitles → Custom Post-Processing → Command:

curl -X POST "http://ai-subtitle-translator:8080/translate" -d input_path="{{subtitles}}";

3. Integration with Existing Bazarr

If you already have Bazarr running, add the translator to your existing docker-compose:

services:
  subtitle-translator:
    build: .
    ports:
      - "8000:8080"
    environment:
      - DEEPSEEK_API_KEY=${DEEPSEEK_API_KEY}
      - PUID=${PUID:-1000}
      - PGID=${PGID:-1000}
    volumes:
      - /path/to/media:/media/subtitles # Same mount as Bazarr
    restart: unless-stopped
    networks:
      - media-network

Alternative Setup Methods

Local Development

1. Install uv:

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

2. Install dependencies:

   uv sync --all-extras
   

3. Set environment variables:

   export DEEPSEEK_API_KEY=your_api_key_here
   

4. Run the API:

   uv run fastapi dev src/ai_subtitle_translator/app.py
   

CLI Usage

Translation:

uv run ai-subtitle-translator 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)

Translation Example:

uv run ai-subtitle-translator movie.srt \
  --translation-mode bilingual \
  --provider deepseek

Extraction from Bilingual Files:

# Extract monolingual subtitles from existing bilingual ASS files
uv run ai-subtitle-translator movie.en-zh.ass --extract-monolingual

This extracts only the translated text with proper monolingual styling and positioning, perfect when you want monolingual versions without re-translating.

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 (For Bazarr Integration)

Process files on the server filesystem using file paths.

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
• translation_mode: bilingual or monolingual - default: bilingual
• prompt_template: full_text or selective_difficulty - default: selective_difficulty

Example:

curl -X POST "http://localhost:8000/translate" \
  -F "input_path=/media/subtitles/movie.srt" \
  -F "provider=deepseek" \
  -F "translation_mode=bilingual"

Response:

{
  "success": true,
  "message": "✅ Translation completed successfully",
  "output_filename": "/media/subtitles/movie.en-zh.ass"
}

Get Available Providers

GET /providers

Returns available AI providers and their configuration status.

Configuration

Environment Variables

Variable	Description	Required
DEEPSEEK_API_KEY	DeepSeek API key (recommended)	Yes
OPENAI_API_KEY	OpenAI API key	For OpenAI provider
GEMINI_API_KEY	Google Gemini API key	For Gemini provider
PUID	User ID for file permissions	Optional (default: 1000)
PGID	Group ID for file permissions	Optional (default: 1000)

Note: Set PUID/PGID to match your host user to avoid permission issues:

# Find your user/group IDs
id

# Set in .env file
PUID=1001
PGID=1001

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 sync --all-extras --dev

Code Quality

# Format code
uv run black .

# Lint code
uv run ruff check .

# Type checking
uv run mypy .

Testing

# Run tests
uv run pytest

# Run with coverage
uv run pytest --cov=src

Health Monitoring

Check service health:

curl http://localhost:8080/health

Check available providers:

curl http://localhost:8080/providers

Performance

• Translation Speed: ~5-30 seconds per file (depends on length)
• API Costs: ~$0.001-0.01 per subtitle file with DeepSeek
• Batch Processing: Optimized for efficiency

Docker Build

Build locally

docker build -t ai-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