How can I insert a time-break at the end of a line/paragraph when using a multilingual voice in Azure?

Question

Hello,
I am using the Azure AI text-to-speech web interface.

When I generate an audio from the following text, both Strong-breaks are generated.

[Guy] This is a text. [Strong] This is a text after a break. [Strong]
[Guy] And this is a text in a new line, after a break at the end of the previous line.

When I generate the audio from the same text, but with another voice, only the first Strong-break is generated, the second Strong-break is ignored.

[Florian Multilingual] This is a text. [Strong] This is a text after a break. [Strong]
[Florian Multilingual] And this is a text in a new line, after a break at the end of the previous line.

All Multiligual voices seem to ignore breaks at the end of a line/paragraph.
With all standard voices the breaks work.

How can I insert a break at the end of a line/paragraph when using a multilingual voice?

Thanks in advance
Martin

Answer 1

I have just found a workaround.

When I add a space after the time-break at the end of the line, the time-break is recognized also with multilingual voices. The result is fine, but you cannot see in the source text, whether there is a space or not. Therefore, it is not a solution, but only a workaround.
It would be nice if Microsoft could fix this problem with a real solution.

Answer 2

I am not expert in this matter but here is what I found on some blogs how you need to structure your SSML to insert breaks, including those at the end of a line or paragraph:

<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xml:lang="en-US">
  <voice name="FlorianMultilingual">
    This is a text. 
    <break strength="strong"/> 
    This is a text after a break.
    <break strength="strong"/>
  </voice>
  <voice name="FlorianMultilingual">
    And this is a text in a new line, after a break at the end of the previous line.
  </voice>
</speak>

or use time in your break tag :

time="2000"

https://stackoverflow.com/questions/75869528/how-to-customize-silence-time-between-sentence-groups-in-azure-text-to-speech