# ToneTwin

> ToneTwin is a language-learning web app that clones your voice from a short recording, then speaks phrases in the language you are learning in your own voice. Hearing yourself say a sentence fluently is the model you practice toward.

Key facts:
- Voice cloning uses the open XTTS v2 model. A voice profile is built from a few English phrases you read aloud.
- Phrases can be spoken in 16 languages: English, Spanish, French, German, Italian, Portuguese, Polish, Turkish, Russian, Dutch, Czech, Arabic, Chinese (Mandarin), Japanese, Hungarian and Korean.
- Listening practice plays a lesson phrase by phrase, hands-free, with speed control and an optional English translation.
- Speaking practice scores your own recording of a phrase from 0 to 100 and marks each word (each character for Chinese and Japanese) as correct, almost right, different or missed. For Mandarin it flags wrong tones. Recognition uses OpenAI's open Whisper model.
- Built-in phrase lessons come with the app; Mandarin Chinese has the most.
- You can import Anki decks (.apkg) as custom lessons, or write your own.
- The Voice Playground translates English to Chinese, shows pinyin, and speaks the result in your voice.
- The home page has a demo: record one English sentence without an account and hear it in another language.
- Everything except the demo needs a free account (email magic link or Google sign-in). Voice recordings and lessons are private to each account.
- ToneTwin has no public API for agents.

## Pages

- [Home](https://tonetwin.app/): what ToneTwin does, plus the try-it-now voice demo
- [Sign in](https://tonetwin.app/login): create an account or sign in