---
title: Audio8 TTS Preview
emoji: 🗣️
colorFrom: green
colorTo: blue
sdk: gradio
sdk_version: 5.50.0
app_file: app.py
python_version: "3.12"
short_description: Zero-shot voice cloning TTS gallery for Audio8 0.6B
startup_duration_timeout: 30m
---

# Audio8 TTS Preview 0.6B

Demo for [Audio8/Audio8-TTS-Preview-0.6b](https://huggingface.co/Audio8/Audio8-TTS-Preview-0.6b),
a 0.6B-parameter DualAR multilingual TTS model with zero-shot voice cloning,
running on ZeroGPU.

Browse thousands of reference voices (sourced from
[Daankular/DramaboxTTS](https://huggingface.co/spaces/Daankular/DramaboxTTS)'s
`voices.json`), pick one to clone, or upload/record your own reference clip.
A CPU-side Whisper pass auto-transcribes the reference clip since Audio8 TTS
requires a matching transcript to condition cloning.

Supported generation languages: Cantonese, Chinese, Dutch, English, French,
German, Italian, Japanese, Korean, Polish, Spanish.