Sirene

Appearance

SireneMulti-backend TTS Router

A single entry point to generate speech via Kokoro, Qwen3-TTS, F5-TTS, Piper, and more — with custom voice management via audio samples.

Get Started
API Reference

Multi-Backend

Route text-to-speech requests to Kokoro, Qwen3-TTS, F5-TTS, Piper, CosyVoice, OpenAudio, or Chatterbox from a single API.

Custom Voices

Create custom voices by uploading audio samples. Automatic transcription via Whisper, waveform preview, and zero-shot voice cloning.

Model Management

Download TTS models on demand from the web interface. Models are lazy-loaded into GPU memory with LRU eviction.

Real-time Updates

Track model downloads and generation progress in real-time via Server-Sent Events powered by PocketBase.

Self-Hosted

Run everything locally with Docker. All services (Nginx, Hono, FastAPI, PocketBase) bundled in a single container.

Open Source

MIT licensed. Built with Bun, Hono, FastAPI, React, and PocketBase.