This issue is currently blocked by microsoft/cognitive-services-speech-sdk-js#96.

Feature Request

Some browsers may not select the correct audio input source for speech recognition. For Cognitive Services, it support alternative audio input source by specifying the device ID (navigator.mediaDevices.enumerateDevices).

Implementation details

• Passing custom audioConfig is already supported in the Cognitive Services adapter
  • (No code modifications needed, just referencing them here)
  • https://github.com/compulim/web-speech-cognitive-services/blob/master/packages/component/src/SpeechServices/SpeechToText/createSpeechRecognitionPonyfill.js#L92
  • https://github.com/compulim/web-speech-cognitive-services/blob/master/packages/component/src/SpeechServices.js#L17
• Building the Web Speech ponyfill factory with specific device
  • Modify https://github.com/microsoft/BotFramework-WebChat/blob/master/packages/bundle/src/createCognitiveServicesSpeechServicesPonyfillFactory.js#L18
  • Around line 19, you will need to pass your audioConfig object. And it will hit SpeechServices.js:17, then hit createSpeechRecognitionPonyfill:92
  • Modify line 4, so the host of Web Chat can specify an audioConfig object
• Build a sample to show capability to select different audioConfig on-the-fly
  • Use AudioConfig.fromMicrophoneInput(deviceId) to create the AudioConfig object with a specific device, https://github.com/microsoft/cognitive-services-speech-sdk-js/blob/master/src/sdk/Audio/AudioConfig.ts#L38

[Enhancement]