yanacuti1121 VoxCPM2: Tokenizer-Free TTS cho multilingual speech generation, creative voice design, và voice cloning.
Install
pip install voxcpm
Quick start
from voxcpm import VoxCPM
model = VoxCPM.from_pretrained("openbmb/VoxCPM2")
# Text to speech
audio = model.synthesize("Hello, world!", lang="en")
# Voice cloning
audio = model.clone_voice(reference_audio="speaker.wav", text="Your text here")
# Multilingual
audio = model.synthesize("Xin chào", lang="vi")
Features
- Tokenizer-free architecture — faster inference
- 20+ languages including Vietnamese
- Voice cloning từ reference audio
- Creative voice design (style control)
Source
https://github.com/OpenBMB/VoxCPM · ⭐25.5K https://huggingface.co/openbmb/VoxCPM2