Skip to content

sherpa-onnx

Run speech recognition and synthesis across platforms

Visit official site

What sherpa-onnx is

sherpa-onnx is described as a speech recognition and speech synthesis project built on next-generation Kaldi, with inference carried out through onnxruntime. The material lists embedded systems, Android, iOS, Raspberry Pi, RISC-V and servers among the supported platforms.

What you can do with it

  • Run streaming and non-streaming speech recognition
  • Convert text into speech
  • Identify or verify speakers
  • Detect audio tags and spoken keywords

Application scenarios

  • Carry out real-time speech to text on an Android device
  • Process batch speech recognition jobs on a server
  • Spot keywords inside an embedded system

Getting started

  • Clone or download the project to a local machine
  • Choose an API and platform for the needed function
  • Configure the environment and dependencies from the docs
  • Load a pre-trained model and run a test

Pros & cons

✓ What we like

  • Inference runs through onnxruntime
  • Many platforms and operating systems are listed

! What to watch out for

  • Inference depends on the onnxruntime setup
  • A pre-trained model is loaded before testing

FAQ

What is it based on?

The material describes it as based on next-generation Kaldi with onnxruntime for inference.

Which platforms are named?

Embedded systems, Android, iOS, Raspberry Pi, RISC-V and servers are listed in the material.

Last reviewed: 2026-11-15

More AI audio tools tools

View all →

How we review