Show HN: Chickadee – Free, local, open-source text-to-speech for web pages

Chickadee is a free, open-source Chrome extension that converts web articles to speech using a voice model that runs locally in your browser, eliminating the need for subscriptions or API keys. The tool, shown on Hacker News by creator Sahil Mahendrakar, uses Kokoro-82M, an 82-million-parameter speech model, and processes audio entirely on your computer via WebGPU.

The extension includes 12 voices, adjustable playback speed, sentence highlighting synchronized to the original page, and requires no account or API key. After an initial model download, the tool works fully offline, meaning users can continue generating speech without an internet connection. The project is open source and available at usechickadee.com.

Mahendrakar built Chickadee to address frustrations with existing text-to-speech options — robotic-sounding voices and the friction of subscriptions or authentication. Running the model locally in the browser solves both problems, though it does require the initial one-time download of Kokoro-82M to the user's device.

The approach reflects growing interest in bringing AI inference to the client side. By avoiding cloud APIs, Chickadee keeps data on the user's machine and eliminates per-use costs or rate limits that plague subscription-based TTS services. Whether the 82-million-parameter model delivers voice quality comparable to larger server-side alternatives remains to be seen, but the trade-off — local privacy and zero cost for offline operation — appeals to users who prioritize those properties over production-grade voice synthesis.