audio.cpp puts 12 audio AI model families in one C++ runtime
audio.cpp is an open-source project for running many audio AI models in one C++ runtime. The repository currently has 12 released model families that are ready for normal use, while models still being connected or optimized are not counted as released. The released models cover , , voice design, , alignment, , voice conversion, audio codecs, and editing.
The named models include Chatterbox, MioTTS, OmniVoice, PocketTTS, Qwen3-TTS, VoxCPM2, Qwen3-ASR, Qwen3 Forced Aligner, Silero VAD, Seed-VC, MioCodec, and Vevo2. The goal is to avoid treating every audio model as a separate setup with its own Python environment, , , batching logic, and deployment process. The shared pieces are meant to include one runtime, session handling, , server, audio utilities, and later higher-level workflows.
According to the title, some work can run up to 5 times faster than Python on CUDA.
Key points
- audio.cpp runs audio AI models through a C++ runtime.
- 12 model families are released and ready for normal use.
- The covered tasks include , , , voice conversion, and editing.
- The project aims to replace many separate Python setups with one shared runtime and server.
- Some work is claimed to be up to 5 times faster than Python on CUDA.