OpenBMB / VoxCPM2
VoxCPM2
Multilingual speech and controllable voice cloning
Windows only
Parameters
2B
Languages
30
Output
48 kHz
Platform
Windows only
Model overview
Built into OpenVox for private local voice generation.
VoxCPM2 is a 2B-parameter tokenizer-free TTS model for multilingual speech, voice design, and controllable reference voice cloning.
VoxCPM2 supports 30 languages and several Chinese dialects without requiring a language tag.
It supports voice design from a text description and controllable cloning from a permitted reference recording.
AudioVAE V2 accepts 16 kHz reference audio and produces 48 kHz output through built-in super-resolution.
In OpenVox, VoxCPM2 is available on Windows and runs locally after its model files are downloaded.
OpenVox does not provide celebrity voice models and does not permit cloning, impersonating, or commercially using any person's voice without proper rights or consent.
Open-source model: OpenBMB / VoxCPM2
View on Hugging Face