Show HN: Kokoro TTS in the browser, 41 voices in 6 languages, runs locally

Category: ai-ml

Tags: text-to-speech, browser-based, local-inference, kokoro, webassembly, audio-synthesis

Score: 5.5/10 (Innovation: 5, Technical: 7, Documentation: 5, Utility: 5)

This project is a web-based implementation of the Kokoro-82M text-to-speech model that runs entirely in the browser, offering 41 voices across 6 languages without requiring server-side processing. It supports text input and document import (EPUB, PDF, etc.) for read-aloud functionality, with all processing done locally for privacy and offline use.

Target audience: end users, developers interested in client-side ML, accessibility advocates

Repository: https://sandbook.app/kokoro-tts-online · C++ · MIT · 28883 stars

View on Hacker News