How to select the voice using GSS in SSML

[Sylvain Gagnon]

Hello,

I try using a voicexml document with ssml prompt how I specified the voice to be used by Google Speech Synthesis plugin ressource? According the Unimrcpserver log I get:

ANNOUNCE rtsp://10.1.1.144:1554/media/speechsynthesizer RTSP/1.0

CSeq: 76359

Content-Type: application/mrcp

Session: 17f9ded8ff244242

Content-Length: 448

SPEAK 119769 MRCP/1.0

Speech-Language: en-US

Content-Type: application/synthesis+ssml

Content-Length: 336

<?xml version="1.0" encoding="utf-8"?>

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

 <voice xml:lang="en-US" name="en-US-Wavenet-C">I hear this is a test of a second message , with a confidence level of 0.00 , which give an interpretation of this is a test of a second message .</voice></speak>

but the voice still the default one "en-US-Wavenet-A", the same problem when I try the "gender" property of the voice tag in SSML.

Thanks,

Sylvain

[Arsen Chaloyan]

Hello Sylvain,

The voice can be specified from the MRCP client by means of the header field Voice-Name. See section 5.4 in the Usage Guide.

This may not be under your control, though, since the VoiceXML platform that you use sets the name as an attribute to <speak> in SSML, which Google Text-To-Speech API does not seem to observe.

If you do not have any means to set the corresponding MRCP header field, then in next version of the plugin, we can add a global configuration parameter in umsgss.xml allowing to specify the default voice name. Alternatively, this issue can be addressed by rewriting the content of <speak> element to make it in compliance with the Text-To-Speech service, which would be harder to achieve but is somewhat already implemented in the BingSS plugin.

--
You received this message because you are subscribed to the Google Groups "UniMRCP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to unimrcp+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Arsen Chaloyan
Author of UniMRCP
http://www.unimrcp.org

[Sylvain Gagnon]

Thanks Arsen for your response, effectively I don't have any control on the custom header of the <speak> in own VoiceXML platform, having a global configuration parameter in umsgss.xml will be great.

To unsubscribe from this group and stop receiving emails from it, send an email to unimrcp+u...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Arsen Chaloyan]

Both voice name and gender can now optionally be specified from umsgss.xml in the development version of the GSS plugin. This change will be available in next public release. In the meantime, if anyone is interested in trying the development version, please contact serv...@unimrcp.org with such a request.

To unsubscribe from this group and stop receiving emails from it, send an email to unimrcp+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Arsen Chaloyan]

For those tracking this discussion, you may have noticed that the new GSS release has been published.

In addition to what discussed here, you may set the configuration parameter bypass-ssml to false in order for the voice parameters specified in SSML content to take effect.

See Section 5.4 for all the supported methods on how to specify voice parameters.

https://unimrcp.org/manuals/pdf/GoogleSSUsageManual.pdf