I've been looking at this also, but on IOS.
User Media uses the HTML5 text to speech API. Example:
var utterance = new SpeechSynthesisUtterance('Hello world');
It doesn't produce an error, and debug shows that window.speechSynthesis is an object. But the call to .speak doesn't produce sound. I attached it to a user interaction (e.g. click a button to .speech) in case it required user interactivity, still no joy.
I tried using the Cordova TTS plugin instead but it clashed with the Construct 2 audio system.
Once I've looked into this more I'll update this post with my findings.