Screenshots

Version

master

Describe the bug

When creating a ponyfill with authorizationToken and region as a string, it is throwing an error saying, "web-speech-cognitive-services: Authorization token must be a string."

This is because the createCognitiveServicesSpeechServicesPonyfillFactory.js is not resolving Promise.

The fix would be:

Steps to reproduce

1. Use authorizationToken (string) and region (string) to create a Cognitive Services ponyfill factory (see the code below)
2. Click on the microphone button

const webSpeechPonyfillFactory = await window.WebChat.createCognitiveServicesSpeechServicesPonyfillFactory( {
  authorizationToken,
  region
} );

Expected behavior

It should start speech recognition.

Additional context

Related to recent PR #2759.

Also, could we allow no reference grammar ID to be passed to the ponyfill?

Also, the error message is also not very visible until putting a breakpoint. Could we improve this?

[Bug]