What's New
🎯 Multi-Provider TTS Support
Choose your preferred TTS provider:
- OpenAI - Original provider
- Deepinfra - Open-source models (Kokoro-82M, Orpheus-3B, Sesame-1B)
- Custom OpenAI-Compatible - Self-hosted or third-party endpoints
Provider-specific voice management with automatic restoration per model.
📦 Package Manager Migration
Migrated to pnpm for improved dependency handling and reproducible builds.
🐳 Docker Improvements
Updated Docker configuration with Deepinfra API defaults and added docker-compose example with Kokoro-FastAPI integration.
🔧 Bug Fixes & Improvements
- Documentation updates
- CI/CD workflow improvements for pnpm
⚠️ Breaking Changes
Voice selection is now provider-model specific. Previously saved voices may need to be reselected.