Download
Phonon TTS is an open-source text-to-speech inference engine that generates natural-sounding speech locally on computers, mobile devices, and web browsers. Built in Rust, it operates without PyTorch or a cloud connection after downloading its models. The software provides compact 31-million and 40-million parameter models for efficient deployment. It supports English, French, German, Spanish, and Portuguese speech synthesis. Users can select predefined voices, create custom voices, or clone voices for supported applications. Its streaming architecture produces audio progressively to minimize playback delays. Hardware acceleration includes Apple Core ML, compatible Snapdragon NPUs, and GPU backends. Phonon integrates through Python, Swift, Rust, JavaScript, Android libraries, and OpenAI-compatible speech APIs.
Features
- Offline Neural Text-to-Speech Synthesis
- Five-Language Speech Generation
- Compact 31M and 40M Parameter Models
- Voice Selection, Creation, and Cloning
- Low-Latency Streaming Audio Generation
- Cross-Platform Hardware-Accelerated Inference
Project Samples
