Just need to speak a text string from the ViewModel (inherited from MVVM Light ViewModelBase) to the MediaElement on the XAML page.
var synthesisStream = await synthesizer.SynthesizeSsmlToStreamAsync(text);
media.AutoPlay = true;
media.SetSource(synthesisStream, synthesisStream.ContentType);
media.Play();
The code above has no separation of ViewModel. We see media is directly handled in code-behind.
In my ViewModel, I stopped at
var synthesisStream = await synthesizer.SynthesizeSsmlToStreamAsync(text);
var msg=new PlaySpeechSynthesisStreamMessage(synthesisStream);
Messenger.Default.Send<PlaySpeechSynthesisStreamMessage>(msg);
For the message:
public class PlaySpeechSynthesisStreamMessage
{
    public SpeechSynthesisStream Stream { get; set; }
    public PlaySpeechSynthesisStreamMessage(SpeechSynthesisStream stream)
    {
        Stream = stream;
    }
}
Is Messenger the right way to handle this situation? How can we write a RelayCommand or something to pass the stream to media?
A related article MVVM pattern violation: MediaElement.Play() seems to address this issue, but it is not in MVVM Light and there is no way to pass the stream, either.
 
     
    