Quick Start

Generate your first speech with Chatterbox TTS

This guide walks you through generating speech with all three Chatterbox models: Turbo, Original, and Multilingual.

Choose Your Model

Turbo (Recommended)

Chatterbox-Turbo is the fastest and most efficient model, perfect for voice agents and production use. It features:

• 350M parameters for lower VRAM usage
• Native paralinguistic tags ([laugh], [chuckle], [cough])
• Single-step mel decoder for ultra-fast generation
• English only

Original

Chatterbox is the original model with creative controls for expressive speech:

• 500M parameters
• Classifier-free guidance (CFG) tuning
• Exaggeration controls
• English only

Multilingual

Chatterbox-Multilingual supports 23+ languages with zero-shot voice cloning:

• 500M parameters
• Supports 23+ languages including Arabic, Chinese, French, German, Spanish, and more
• Zero-shot voice cloning across languages

Generate Your First Audio

1. Import the library

   Import the necessary modules based on which model you want to use:

   Turbo

   import torchaudio as ta
   import torch
   from chatterbox.tts_turbo import ChatterboxTurboTTS
   

   import torchaudio as ta
   import torch
   from chatterbox.tts import ChatterboxTTS
   

   import torchaudio as ta
   import torch
   from chatterbox.mtl_tts import ChatterboxMultilingualTTS
   
2. Load the model

   Initialize the model with automatic device detection:

   Turbo

   # Load the Turbo model
   model = ChatterboxTurboTTS.from_pretrained(device="cuda")
   

   # Automatically detect the best available device
   if torch.cuda.is_available():
       device = "cuda"
   elif torch.backends.mps.is_available():
       device = "mps"
   else:
       device = "cpu"
   
   model = ChatterboxTTS.from_pretrained(device=device)
   

   # Automatically detect the best available device
   if torch.cuda.is_available():
       device = "cuda"
   elif torch.backends.mps.is_available():
       device = "mps"
   else:
       device = "cpu"
   
   model = ChatterboxMultilingualTTS.from_pretrained(device=device)
   
3. Generate speech

   Create audio from text:

   Turbo

   # Generate with Paralinguistic Tags
   text = "Oh, that's hilarious! [chuckle] Um anyway, we do have a new model in store."
   
   # Generate audio
   wav = model.generate(text)
   

   # English text
   text = "Ezreal and Jinx teamed up with Ahri, Yasuo, and Teemo to take down the enemy's Nexus in an epic late-game pentakill."
   
   # Generate audio
   wav = model.generate(text)
   

   # French text
   text = "Bonjour, comment ça va? Ceci est le modèle de synthèse vocale multilingue Chatterbox, il prend en charge 23 langues."
   
   # Generate audio with language ID
   wav = model.generate(text, language_id="fr")
   

   Turbo supports paralinguistic tags like [laugh], [chuckle], [cough] to add natural expressions to speech.

4. Save the audio

   Save the generated audio to a file:

   Turbo

   ta.save("output-turbo.wav", wav, model.sr)
   

   ta.save("output-english.wav", wav, model.sr)
   

   ta.save("output-french.wav", wav, model.sr)
   

Voice Cloning

All models support zero-shot voice cloning using a reference audio file. Provide an audio prompt to clone any voice:

Turbo

# Generate with voice cloning
text = "Hi there, Sarah here from MochaFone calling you back [chuckle]"
wav = model.generate(text, audio_prompt_path="your_10s_ref_clip.wav")
ta.save("cloned-voice.wav", wav, model.sr)

# Generate with voice cloning
text = "This will sound like the reference speaker"
wav = model.generate(text, audio_prompt_path="YOUR_FILE.wav")
ta.save("cloned-voice.wav", wav, model.sr)

# Generate with voice cloning in Chinese
text = "你好，今天天气真不错，希望你有一个愉快的周末。"
wav = model.generate(text, audio_prompt_path="YOUR_FILE.wav", language_id="zh")
ta.save("cloned-voice-chinese.wav", wav, model.sr)

For best results, use a reference audio clip that is 5-10 seconds long with clear speech and minimal background noise.

Complete Examples

Here are complete working examples for each model:

Turbo

import torchaudio as ta
import torch
from chatterbox.tts_turbo import ChatterboxTurboTTS

# Load the Turbo model
model = ChatterboxTurboTTS.from_pretrained(device="cuda")

# Generate with Paralinguistic Tags
text = "Oh, that's hilarious! [chuckle] Um anyway, we do have a new model in store. It's the SkyNet T-800 series and it's got basically everything. Including AI integration with ChatGPT and all that jazz. Would you like me to get some prices for you?"

# Generate audio
wav = model.generate(text)
ta.save("test-turbo.wav", wav, model.sr)

import torchaudio as ta
import torch
from chatterbox.tts import ChatterboxTTS

# Automatically detect the best available device
if torch.cuda.is_available():
    device = "cuda"
elif torch.backends.mps.is_available():
    device = "mps"
else:
    device = "cpu"

print(f"Using device: {device}")

model = ChatterboxTTS.from_pretrained(device=device)

text = "Ezreal and Jinx teamed up with Ahri, Yasuo, and Teemo to take down the enemy's Nexus in an epic late-game pentakill."
wav = model.generate(text)
ta.save("test-english.wav", wav, model.sr)

import torchaudio as ta
import torch
from chatterbox.mtl_tts import ChatterboxMultilingualTTS

# Automatically detect the best available device
if torch.cuda.is_available():
    device = "cuda"
elif torch.backends.mps.is_available():
    device = "mps"
else:
    device = "cpu"

model = ChatterboxMultilingualTTS.from_pretrained(device=device)

# French example
french_text = "Bonjour, comment ça va? Ceci est le modèle de synthèse vocale multilingue Chatterbox, il prend en charge 23 langues."
wav_french = model.generate(french_text, language_id="fr")
ta.save("test-french.wav", wav_french, model.sr)

# Chinese example
chinese_text = "你好，今天天气真不错，希望你有一个愉快的周末。"
wav_chinese = model.generate(chinese_text, language_id="zh")
ta.save("test-chinese.wav", wav_chinese, model.sr)

Supported Languages (Multilingual Model)

The Multilingual model supports 23+ languages:

Arabic (ar) • Danish (da) • German (de) • Greek (el) • English (en) • Spanish (es) • Finnish (fi) • French (fr) • Hebrew (he) • Hindi (hi) • Italian (it) • Japanese (ja) • Korean (ko) • Malay (ms) • Dutch (nl) • Norwegian (no) • Polish (pl) • Portuguese (pt) • Russian (ru) • Swedish (sv) • Swahili (sw) • Turkish (tr) • Chinese (zh)

Next Steps

• Learn about advanced features and tuning parameters
• Explore paralinguistic tags for Turbo
• Understand voice cloning best practices
• Check out the watermarking features for responsible AI