Get started with Speech Service client libraries

Besides making direct HTTP requests via a REST API, Speech Service provides developers with Speech client libraries in different languages. The Speech client libraries:

  • Support more advanced capabilities in speech recognition, such as intermediate results in real time, long audio stream (up to 10 minutes), and continuous recognition.
  • Provide a simple and idiomatic API in the language of your preference.
  • Hide low-level communication details.

Currently, the following Speech client libraries are available:

Additional resources

  • The samples page provides complete samples to use Speech client libraries.
  • If you need a client library that's not yet supported, you can create your own SDK. Implement the Speech WebSocket protocol on the platform and use the language of your choice.


All Cognitive Services SDKs and samples are licensed with the MIT License. For more information, see License.