CoolFace
Apppublic

slam25/chattts-studio

sourceHugging Facemitupdated 5mo agoView on Hugging Face
0likes
App README

🎙️ ChatTTS Studio

A Gradio web interface for ChatTTS — a 300M-parameter conversational text-to-speech model optimized for natural, expressive dialogue.

Features

  • —Real-time audio playback — auto-plays generated speech in the browser
  • —Speed control — 9 speed levels (1 = slowest, 9 = fastest)
  • —Prosody controls — oral/conversational tone, laugh tendency, pause length
  • —Temperature — controls voice variation and diversity
  • —Random speaker sampling — sample a new voice identity at any time
  • —Chat history UI — tracks all synthesized turns
  • —Inline text controls — embed [uv_break], [laugh], [lbreak] directly in text

Quick Start

  1. 1.Type or paste text into the input box
  2. 2.Adjust Speed, Temperature, and Prosody sliders
  3. 3.Click ▶ Generate Speech or press Enter
  4. 4.Audio auto-plays; use the player to replay or download

Text Control Tags

TagEffect
[uv_break]Short micro-pause
[lbreak]Longer sentence break
[laugh]Laughter

Model

Built on 2Noise/ChatTTS.

Notes

  • —First inference takes 30–60 seconds for model warm-up on CPU
  • —Speed 4–6 sounds most natural for conversational content