TTS Voices: Do Microsoft's Voices Support Alpha Tags?

Question

I wonder, do Microsoft's voices like Christopher, Jenny, Aria, Guy, and so on, support Alpha tags? Things like [announcing], [friendly], [chuckles], and the like?

Answer 1

Hello. Microsoft’s built-in voices such as Christopher, Jenny, Aria, Guy, and others do not support “alpha tags” like [chuckles], [announcing], or [friendly]. Those kinds of inline cues aren’t recognized. What they do support is SSML (Speech Synthesis Markup Language) and Microsoft’s own SSML extensions, which let you adjust tone, style, pauses, and other prosody features. For example, you can use <mstts:express-as style="friendly"> or insert an <audio> clip if you want a laugh or sound effect.

You can find the full list of supported tags and styles here:

• SSML reference for Speech Service: https://learn.microsoft.com/en-us/azure/ai-services/speech-service/speech-synthesis-markup
• Microsoft SSML extensions (express-as, role, emotions): https://learn.microsoft.com/en-us/azure/ai-services/speech-service/speech-synthesis-markup-structure

So while you can’t drop in “[chuckles]” directly, you can achieve similar results with SSML styles or by embedding audio.

Answer 2

Exactly! That will work, but the <mstts:express-as> tag needs to be opened properly and then closed around the text. Right now, you only have a closing tag </mstts:express-as> without the opening part, so it won’t work. A valid structure looks like this:

<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis"
       xmlns:mstts="https://www.w3.org/2001/mstts"
       xml:lang="en-US">

  <voice name="en-US-JennyNeural">
    <mstts:express-as style="friendly">
      Welcome aboard United Airlines, Flight 1697, with nonstop service to San Francisco.
      My name is Jenny, and I am your chief flight attendant this morning. 
      We would like your undivided attention as we demonstrate the safety features 
      of this Canadair RJ 700 aircraft.
    </mstts:express-as>
  </voice>

</speak>

• Always open and close <mstts:express-as> around the text you want styled.
• The style attribute must be one supported by the voice (e.g. friendly, chat, customerservice).
• Don’t put quotes around the spoken text inside SSML. Just plain text inside the tag.

You can check supported styles and roles for each voice here: https://learn.microsoft.com/en-us/azure/ai-services/speech-service/speech-synthesis-markup-structure

Answer 3

Thanks for the update. In this case, the Voice Gallery page you used isn’t built to process SSML. It’s just a simple demo that plays plain text using preset styles. So when you pasted in your full markup, it didn’t recognize the tags, it just read them out loud like regular text.

To properly test SSML (like <mstts:express-as>), you’ll want to switch to the Text-to-Speech tool inside Azure Speech Studio:

1. Go to https://speech.microsoft.com/portal
2. Sign in with your Microsoft or Azure account
3. Click “Text to Speech” under Custom Voice
4. Choose Jenny Neural or whichever voice you like
5. Paste your full SSML
6. Click Play

That tool will honor the tags, and Jenny should speak in the style you defined (like “friendly” or “chat”).