Easy-to-use libraries, introductory tutorial, code examples, debugging tools ... an ecosystem for successfull development of voice based applications.

Python Client Library

We developed a Python client library that enables you to seamlessly use Soniox speech recognition in any application. With only a few lines of code you can transcribe files or audio streams in real-time low-latency scenarios.

We use the library internally at Soniox in our services that run in the cloud under heavy load.

Multi-Part Tutorial

We wrote a multi-part tutorial to describe step by step all the concepts of Soniox speech recognition system. Once you are done reading the tutorial, you should have a complete understanding of all the available features and how to use them in your applications.

The tutorial teaches you how to transcribe files or any audio stream under different application requirements. It shows you to capture and transcribe audio from your microphone. The tutorial also demonstrates how to use speech adaptation to bias the recognition towards ceratin words, phrases or domains.

Start tutorial


We wrote two examples of real-world applications: Transcribe Files in Parallel and Transcribe Multiple Streams that demonstrate how to use Soniox in production environments and do large scale real-time low-latency speech recognition.

Soniox Cloud

We built Soniox Cloud web application so anyone can seamlessly transcribe live audio from the microphone or transcribe uploaded files. It’s a great tool for testing and debugging the output of Soniox speech recognition system.

Under My Transcripts tab, you can review all of your speech recognition requests individually. It enables you to quickly evaluate the output of Soniox speech recognition and compare it with your expected output.

You can also manage your API keys which you will need to access the Soniox API.

Developer Support

Developers can sometimes get stuck and need help to move forward. We use GitHub to track all of the ideas, issues, requests and bugs related to Soniox speech recognition.

We are super fast in responding to issues and requests. You should receive the response within 24 hours. We also encourage other developers to contribute and engage in raised issues and requests.