How to use from the
Use from the
MLX library
# Download the model from the Hub
pip install huggingface_hub[hf_xet]

hf download starkdmi/chatterbox --local-dir chatterbox

Chatterbox Multilingual (MLX)

Multilingual Chatterbox TTS weights in MLX format (fp32).

Files

  • model.safetensors (2.53 GB)
  • config.json
  • grapheme_mtl_merged_expanded_v1.json
  • Cangjie5_TC.json

Usage (Swift)

generate \
  --weights-dir /path/to/weights \
  --text "Hello world." \
  --lang en \
  --ref-wav-path /path/to/reference.wav \
  --out-wav /path/to/output.wav

Usage (Python MLX)

from chatterbox.model import Model

model = Model.from_pretrained("/path/to/weights")

Notes:

  • S3TokenizerV2 weights are loaded from the HF cache (mlx-community/S3TokenizerV2) unless included.
Downloads last month
21
Safetensors
Model size
0.7B params
Tensor type
F32
ยท
MLX
Hardware compatibility
Log In to add your hardware

Quantized

Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support