Kokoro TTS Engine

Privacy-friendly, offline, 82M param, open-weight text-to-speech (TTS) engine

The "Kokoro TTS Engine" browser extension adds a few new natural-sounding English voices to your browser. Powered by an 82M-parameter large language model (LLM), it converts text to speech fully offline, ensuring privacy—your text never leaves your device. Kokoro has no standalone interface. Its voices are appended to the browser's Speech Synthesis API, so any webpage or extension using this API can access them automatically. The extension downloads required model data only once and stores it in the browser cache. Changing precision settings may trigger new downloads, but all data remains local. Use the Factory Reset button to clear redundant cache if needed. Kokoro supports CPU, GPU (WebGPU), and WASM engines. GPU mode offers the best performance; WASM works on browsers without WebGPU. TTS is CPU-intensive, so high-performance devices are recommended. Adjust engines or data precision to optimize speed and quality. When first fetching resources, a progress window shows download status and URLs. Keep it open until completion; it closes automatically. Features: Fully offline, no server data transmission High-quality, natural-sounding voices Secure, on-device processing Seamless integration via Speech Synthesis API CPU, GPU, and WASM support Adjustable precision for performance optimization This extension brings advanced, private, and natural-sounding text-to-speech to your browser—ideal for anyone who wants high-quality voice output without relying on external servers.