Is there GRPC support for Speech to Text in Azure Speech SDK in java?

Question

Hi,

Is there GRPC support for Azure speech SDK? We are looking for this support for the Realtime Speech to Text feature. Is that support available in Java?

If there is no GRPC support, what is the underlying architecture, and how is the voice streamed to the model? I would appreciate any documentation links on the architecture.

Thanks, Sai Vishnu Soudri

Answer 1

Hi Sai Vishnu Soudri,

Thanks for Reaching the Microsoft Q&A Forum.

The Azure Speech SDK does not currently support GRPC for its real-time Speech-to-Text feature. Instead, it uses Web Sockets to stream audio data in real-time to the Azure Speech Service. This protocol enables efficient, low-latency, bidirectional communication, allowing the user to send audio and receive transcriptions as they are processed. The underlying architecture for Azure Speech-to-Text relies on capturing live audio via the user-side Speech SDK and streaming it to Azure’s Speech Service using Web Sockets Once the audio reaches Azure, it is processed by deep learning-based speech recognition models, which transcribe spoken language into text. The SDK handles session management, retries, and error reporting, ensuring a robust and seamless real-time transcription experience.  For further details, you can explore the Azure Speech SDK Overview.

Thank you!