README.md

metadata

library_name: mlx-audio-plus
base_model:
  - ResembleAI/chatterbox
tags:
  - mlx
  - multilingual
  - tts
  - text-to-speech
  - japanese
language:
  - ar
  - da
  - de
  - el
  - en
  - es
  - fi
  - fr
  - he
  - hi
  - it
  - ja
  - ko
  - ms
  - nl
  - 'no'
  - pl
  - pt
  - ru
  - sv
  - sw
  - tr
  - zh
pipeline_tag: text-to-speech

🚨モデルの完全な動作確認がまだできていません！ (🚨We're still working on fully testing the model!)

YUGOROU/Chatterbox-Multilingual-MLX-4bit

Chatterbox Multilingual TTS converted to MLX format for Apple Silicon devices.

🌍 Supported Languages (23 languages)

Arabic, Danish, German, Greek, English, Spanish, Finnish, French, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Dutch, Norwegian, Polish, Portuguese, Russian, Swedish, Swahili, Turkish, Chinese

📥 Installation

pip install -U mlx-audio-plus

🚀 Usage

Command Line

mlx_audio.tts.generate \\
    --model {model_name} \\
    --text "こんにちは、元気ですか？" \\
    --ref_audio reference.wav

Python

from mlx_audio.tts.generate import generate_audio

generate_audio(
    text="こんにちは、元気ですか？",
    model="{model_name}",
    ref_audio="reference.wav",
    file_prefix="output",
)

📊 Model Details

• Base Model: ResembleAI/chatterbox
• Tokenizer: 2454 tokens (Multilingual)
• Quantization: {'4-bit' if '4bit' in model_name else '8-bit' if '8bit' in model_name else 'fp16'}
• Framework: MLX (Apple Silicon optimized)

🔗 Related

• Original PyTorch model: ResembleAI/chatterbox
• S3Tokenizer: mlx-community/S3TokenizerV2