OpenBMB releases VoxCPM2 open-source multilingual TTS model
OpenBMB announced VoxCPM2, a 2B-parameter tokenizer-free text-to-speech model built on the MiniCPM-4 backbone. According to the release description, it was trained on more than 2 million hours of multilingual speech and supports 30 languages, voice design, controllable voice cloning, and 48kHz output. OpenBMB also released the model weights and code under the Apache-2.0 license, while describing the system as an end-to-end diffusion autoregressive architecture.
VoxCPM2 combines multilingual speech generation, prompt-based voice design, controllable cloning, and commercially usable open weights and code, making it a notable open TTS release.
VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning

VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning
English | 中文