🐱🎙️ New Feature: In-Browser TTS with KittenTTS
Hey Indie Hackers,
I just shipped a new feature for QuickEditVideo.com: in-browser text-to-speech powered by a small but surprisingly capable open model called KittenTTS (15M parameters):
👉 You can try it right now here: QuickEditVideo.com/tts
🔧 Background
When building QuickEditVideo, I often needed quick narration or voiceovers for clips. Most TTS solutions I tried were:
Cloud-only (expensive + slow)
Proprietary (black box models)
Or simply too heavy to run in-browser
So I decided to integrate a lightweight, functional, open TTS model directly into the browser.
✨ What’s new in this release
On top of making KittenTTS run locally, I’ve added some quality-of-life features:
WASM → Turn original Python implementation into JS+WASM, making it fully functional in the browser.
Queueing → Handle multiple text tasks smoothly without blocking main UI thread.
Task history → Reuse and manage your generated audio.
Performance improvements → Optimized for faster inference in-browser.
Text auto-chunking → Splits and sanitizes long text into chunks automatically.
The result: you can paste text, hit generate, and get a voiceover within seconds, no server round trips.
🌟 Why this matters
Runs in your browser — no API keys, no logins, no data sent to servers.
Small model, big usability — 15M parameters sounds tiny compared to giants, but it’s fast and good enough for many voiceover needs.
Open & hackable — I want indie devs and creators to actually see how things work, not just consume from a black box.
🔮 Next steps
More voice presets and styles
Export audio directly into QuickEditVideo’s video timeline
Potential premium voices for those who need studio-level polish
👉 You can try it right now here: QuickEditVideo.com/tts
I’d love feedback from IH:
Would you use in-browser TTS for quick prototyping / voiceovers?
What’s more important to you — speed, voice quality, or cost?
Thanks for reading, and I hope some of you try it out!
