Introducing Polyvox: Clone Any Voice in Seconds
We built Polyvox to make voice cloning accessible and fun. Here's why, and how it works under the hood.
Voice cloning used to be expensive, slow, and locked behind enterprise contracts. We thought that was ridiculous. Everyone should be able to clone a voice and have fun with it, send a birthday message in Japanese, prank a friend in French, or just hear what they'd sound like speaking Korean.
That's why we built Polyvox. Upload or record a 3-second audio sample, and our AI learns the unique characteristics of that voice, tone, pitch, cadence, everything that makes it recognizable. Then type any text, pick from 40+ languages, and get natural-sounding speech that preserves the original voice.
Under the hood, Polyvox uses state-of-the-art neural speech synthesis. The voice embedding is extracted from your sample and combined with a multilingual text-to-speech model that handles pronunciation, intonation, and rhythm for each target language. The result sounds remarkably close to the real thing.
We're launching with a generous free tier, 3 clips, 1 voice clone, and 8 popular languages. Upgrade to Pro for 500 clips/month, 5 clones, and all 40+ languages at just $10/month. Share your clips and earn more for free. No gotchas.
This is just the beginning. We have a lot planned, more voices, longer clips, an API for developers, and features we haven't even announced yet. We'd love for you to try it and tell us what you think.