Multiple Lexicons

Shin Iwamoto 1

Is it possible to use multiple lexicons on Text-To-Speech?
That is, the second lexicon in the following SSML code works?
<speak ....>
<voice name="en-US-JennyNeural">
<lexicon uri="https://..../my-container/sample1.pls"/>
<lexicon uri="https://..../my-container/sample2.pls"/>
text....
</voice>
</speak>

Thanks.

romungi-MSFT 41,861 Reputation points Microsoft Employee

2022-11-16T07:05:12.303+00:00
@Shinlwamoto-5513 I think it is not possible to use two lexicon files with the same voice. If you use two different voices in the XML then you can add two different lexicon uri tags and synthesize them. Here is an example, that should work.

<voice name="en-AU-AnnetteNeural"><lexicon uri="https://cvoiceprodweu.blob.core.windows.net/acc-public-files/file1.xml" />BTW</voice> <voice name="en-US-JennyNeural"><lexicon uri="https://cvoiceprodweu.blob.core.windows.net/acc-public-files/file2.xml" />sdas fdfd</voice></speak>

You could try the scenario using the speech studio Audio content creation tool and test all the features to create a XML that works work for you instead of manually building an XML from scratch.
I hope this helps!! Thanks!!
Shin Iwamoto 1 Reputation point

2022-11-17T00:48:26.16+00:00

Thanks for your answer.
Because a speech text must be in <voice> tag, two voice tags have two different speech.
In your example, if <lexicon> tag is not specified, audio speaking output is "BTW sdas fdfd".
This is not my intention.
Is it SSML standard that using two lexicon files with the same voice is not possible?
Thanks.
romungi-MSFT 41,861 Reputation points Microsoft Employee

2022-11-17T11:50:20.62+00:00

I couldn't find an example from the specification document to define two URLs within the same voice tag.
If you are using the same voice, is it not possible to create multiple lexeme entities in one single file?