k2-fsa/sherpa-onnx
Offline on-device speech AI framework with onnxruntime
- Chart
- Not on a chart right now
- Stars
- 16.1k
What it is
Sherpa-onnx runs speech-to-text, text-to-speech, speaker diarization, and other audio tasks locally without an internet connection using next-gen Kaldi and onnxruntime. It targets embedded systems, mobile devices, desktops, servers, and WebAssembly.
Who it is for
Developers building offline speech and audio features for mobile apps, embedded hardware, or servers would use it.
Why it is trending
It released v2.0.2 on November 19, 2026, and gained 21 stars today. It stands out for supporting 12 programming languages alongside a wide range of edge platforms and NPUs.
Main features
- Streaming and non-streaming speech recognition (ASR) and synthesis (TTS)
- Speaker diarization, keyword spotting, speech enhancement, and VAD
- Runs on Android, iOS, HarmonyOS, Raspberry Pi, RISC-V, and 5 NPU types
- APIs for 12 programming languages plus WebAssembly, Flutter, and Tauri
Summarized on November 19, 2026 from the README and repository data. It can be wrong; check the source before relying on it. Report a wrong summary