KittenTTS WASM

Experimental WebAssembly (WASM) inference app of KittenTTS, a high-quality, lightweight Text-to-Speech engine. This project enables high-performance, private, on-device speech synthesis directly in the browser using ONNX Runtime Web.

Demo : https://mtb0x1.github.io/kittentts_wasm/

Demo with tracing in web console : https://mtb0x1.github.io/kittentts_wasm/?tracing=on

Prerequisites

Before building, ensure you have the following installed:

• Rust (Edition 2024)
• wasm-pack
• Just (optional, but recommended for shortcut commands)
• Python 3 (for serving the web interface locally)

How to Build

Using just (recommended):

just build-release

Or using wasm-pack directly:

wasm-pack build --target web --release --out-dir=web/pkg

This will generate the WASM binary and JavaScript glue code in the web/pkg directory.

How to Use

Using just:

just serve

Or using Python:

python3 -m http.server 8000

Then, open your browser and navigate to: http://localhost:8000/web/index.html

Advanced: Tracing/Debugging

You can enable detailed logging in the browser console by appending a tracing parameter to the URL:

• ?tracing=on: Enables full tracing.
• ?tracing=trace: Sets log level to trace (supports debug, warn, error, etc.).

License

TBD (See Cargo.toml)

---

Note: This is an experimental project and some changes might occur time to time Todo.