[GitHub Trending] OpenBMB/VoxCPM
7.3 relevance
Score Breakdown
technical depth 8
novelty 8
actionability 4
community 7
strategic 5
personal 3
Scored daily by a customisable AI persona to surface the most relevant engineering leadership news.
TTS model, not directly relevant to software engineering infrastructure.
Summary
VoxCPM2 is a tokenizer-free TTS built on a MiniCPM-4 backbone with 2B parameters trained on over 2 million hours of multilingual data, supporting 30 languages and Chinese dialects. It enables voice design from natural-language descriptions, controllable cloning from short clips, and outputs 48kHz audio via AudioVAE V2. Real-time streaming achieves ~0.3 RTF on RTX 4090, accelerated by Nano-vLLM or vLLM-Omni with PagedAttention and an OpenAI-compatible API, all under Apache-2.0.
Author
OpenBMB