We are using Headwind MDM Kiosk Browser 1.29 to display a web application
on Android tablets. The app uses WebRTC with OpenAI Realtime API for voice
interaction (getUserMedia for microphone access).
Problem: When opening the voice feature, getUserMedia throws NotReadableError
- "Could not start audio source". Microphone permission is granted at the
Android level.
This happens specifically on the second open — after the first session closes,
the mic is not fully released by Kiosk Browser, so the next getUserMedia call
fails.
Questions:
1. Does Kiosk Browser 1.29 support microphone access for WebRTC web apps?
2. Is there a known fix or setting to properly release the microphone between sessions?
3. Is there a newer version of Kiosk Browser that fixes this?