Open-source app replicates ElevenLabs voice cloning for free
Original: ElevenLabs just lost its moat 🤯
Source: x.com ↗
Who: Posted by @HowToAI_, an account that curates AI tool discoveries. The subject is Voicebox, an open-source desktop application, author listed as willwillems on GitHub.
What's new: A free, locally-running app called Voicebox can clone any voice from a three-second audio clip, convert text to speech in 23 languages, and apply audio effects — all without sending your data to any outside server. It targets paying users of ElevenLabs (a voice-cloning service at $99 per month) and WisprFlow (a voice-typing tool at $15 per month).
How it works: Voicebox bundles seven separate engines under one interface. A built-in can rewrite text to match a cloned voice's speaking style before the audio is generated. It also includes a so that developer tools can pipe spoken output back to you. Audio effects like reverb and pitch shifting are included. The app is built on rather than the heavier framework, and it runs on , , , and ordinary .
The numbers: The project has collected 23,400 stars on GitHub, a rough measure of community interest. It is released under the , meaning it is completely free to use or build upon.
Caveats: The post is a promotional tweet, not an independent review, so quality comparisons with ElevenLabs are not provided. Voice cloning from three seconds of audio is a known capability of several existing tools; what is new here is the local, free packaging. Ease of setup for non-technical users is not addressed.