AI/ML Speech Synthesis Platform
14.3k 2026-04-18
OpenBMB/VoxCPM
A tokenizer-free, multilingual Text-to-Speech system offering advanced voice design, controllable cloning, and high-quality audio output.
Core Features
30-Language Multilingual Support
Voice Design from natural language descriptions
Controllable and Ultimate Voice Cloning
48kHz High-Quality Audio Output
Real-Time Streaming with acceleration
Detailed Introduction
VoxCPM2 is a cutting-edge, tokenizer-free Text-to-Speech (TTS) system leveraging a 2B parameter diffusion autoregressive architecture. Trained on over 2 million hours of multilingual speech, it supports 30 languages and delivers highly natural, expressive synthesis. Key capabilities include creating new voices from text descriptions, precise voice cloning with style control, and generating studio-quality 48kHz audio. It's fully open-source and optimized for real-time performance, making it suitable for diverse commercial applications.