Azure speechsdk is not working after the deployment of the app in linux based web application (App service)

Question

We have identified a issue with accessing the client’s microphone in our Azure DevOps-deployed web application. The problem arises because the use_default_microphone=True setting in Azure Speech SDK tries to access the server’s microphone instead of the user’s browser microphone. While this works in a local environment, it fails in the cloud since the server does not have a microphone.
The application we are building includes both Speech-to-Text (STT) and Text-to-Speech (TTS) functionalities, and everything must run live with minimal latency to ensure a smooth user experience.

and we are getting this error "exception error Azure Speech to text 0x38 (SPXERR_AUDIO_SYS_LIBRARY_NOT_FOUND)" on linux app service

Answer 1

Hi @Bharani Karthick S,

Thank you for reaching out to Microsoft Q&A forum!

In App Service microphone access is not available because the server lacks physical audio input devices. To fix this, you need to capture audio from the user's browser and send it to your backend for processing. Instead of using the default microphone, use web-based audio recording and then send the recorded audio file on backend. On the server, process the uploaded file using Azure Speech SDK with speechsdk.audio.AudioConfig(filename="YourAudioFile.wav"). Ensure your backend can receive and save audio files, then use the following this code to transcribe.

Also, please refer to: Speaking participant's microphone has a problem.

I hope you understand! Thank you.