OpenBMB has officially launched VoxCPM2, an innovative speech generation model built upon the foundation of MiniCPM-4. This cutting-edge model adopts a tokenizer-free architecture and leverages diffusion autoregressive technology, setting a new benchmark in the field. With its advanced capabilities, VoxCPM2 can generate context-aware speech and perform zero-shot voice cloning with remarkable precision. Moreover, it achieves low-latency streaming synthesis even on consumer-grade GPUs, making it an ideal solution for a wide range of applications, including intelligent voice assistants and audio content creation.
