Hi
I am troubleshooting an issue on AAEP 6.1, Orchestration Designer and Vocalizer for Network as the TTS server.
Same OD application enables you to choose between English or Mexican Spanish TTS. English works successfully, Spanish does not.
The difference between a good call and a bad call is the way the request to the TTS is being made. The requests use different content-type (good call uses plain-text, bad call uses application/synthesis+ssml).
I want to know if this is the way that it should be developed or it may be an issue with Nuance not accepting ssml:
GOOD CALL:
@2012-12-03 12:28:50,905||FINEST|MRCP|29999|FileName=mrcpv1/RtspConnection.cpp,LineNumber=419|~Msg: Sending: ANNOUNCE
rtsp://10.60.188.19/media/speechsynthesizer RTSP/1.0
CSeq: 6
Session: FALMOCLCAAAGPEDNAAAAAAAA
Content-type: application/mrcp
Content-length: 213
SPEAK 1 MRCP/1.0
content-type: text/plain
kill-on-barge-in: true
speech-language: en-US
voice-gender: female
voice-name: Samantha
Content-length: 56
En un lugar de la mancha cuyo nombre no quiero acordarme|DSSVMPP102####
BAD CALL:
@2012-12-03 11:04:51,005||FINEST|MRCP|16172|FileName=mrcpv1/RtspConnection.cpp,LineNumber=419|~Msg: Sending: ANNOUNCE
rtsp://10.60.188.19/media/speechsynthesizer RTSP/1.0
CSeq: 2
Session: FALMMPABAAAMOBCAAAAAAAAA
Content-type: application/mrcp
Content-length: 344
SPEAK 1 MRCP/1.0
content-type: application/synthesis+ssml
kill-on-barge-in: true
speech-language: es-MX
voice-gender: female
voice-name: Paulina
Content-length: 171