Back to Gap Map Model components / Speech & audio

sherpa-onnx

open source / Overall score: 4.0(strong)

The Next-gen Kaldi project's ONNX Runtime engine for offline speech: streaming and non-streaming recognition, synthesis, speaker diarization and identification, voice activity detection and more, on Linux, Windows, macOS, Android, iOS, HarmonyOS, embedded boards and WebAssembly, with bindings for a dozen programming languages.

Openness

5 medium confidence

5.0

license: Apache-2.0(OSI)
source: public
core features withheld: no

sherpa-onnx is Apache-2.0 and the published repository is the whole engine. No maintainer statement about commercial offerings was found, so that reading comes from the repository alone.

https://raw.githubusercontent.com/k2-fsa/sherpa-onnx/HEAD/LICENSE recorded 2026-09-27
LICENSE body: "Apache License" "Version 2.0, January 2004".
https://raw.githubusercontent.com/k2-fsa/sherpa-onnx/HEAD/README.md recorded 2026-09-27
README of the Next-gen Kaldi project; every feature listed builds from the repository, and no paid edition or license key is mentioned.

Adoption

4 high confidence

4.0

PyPI downloads of sherpa-onnx. Its mobile and native builds ship outside PyPI and are not counted.

https://pypistats.org/api/packages/sherpa-onnx/recent recorded 2026-09-27
"last_month":1228049 for sherpa-onnx.

Capability

4 medium confidence

4.0

The broadest on-device speech engine here, serving most speech tasks, but an inference engine rather than a training toolkit.

https://raw.githubusercontent.com/k2-fsa/sherpa-onnx/HEAD/README.md recorded 2026-09-27
"Speech-to-text (i.e., ASR); both streaming and non-streaming are supported"; "Text-to-speech (i.e., TTS)"; "Speaker diarization".

Verified 2026-09-27