Google TTS Standard Names and Neural Names Issue

Ed James

unread,

Oct 13, 2021, 12:26:46 PM10/13/21

to UniMRCP

Hi All,

I'm trying to get TTS working with Google and am sending in SSML which used to work but now doesn't as Google are saying that from 1st September 2021 standard voices are no longer supported for new users (although I don't know why Google thinks I'm a new user) and neural voices are required instead.

In the extract from the uniMRCP server log below I've highlighted the SSML and Google's response. Does anyone know what the equivalent neural voice name would be or does anyone have an example of SSML which works successfully with Google TTS with neural voice name ?

Thanks

Ed James

2021-10-13 16:44:03:811958 [INFO] Receive MRCPv2 Data 172.16.1.179:1544 <-> 10.1.6.80:19413 [354 bytes]

MRCP/2.0 354 SPEAK 1

Channel-Identifier: b0cb59a63bf648fd@speechsynth

Content-Type: application/synthesis+ssml

Content-Length: 217

<?xml version="1.0"?><speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xml:lang="en-GB"><voice name="en-GB-Standard-A">I have of late wherefore I know not lost all my myrth twenty</voice></speak>

2021-10-13 16:44:03:812339 [INFO] Assign Control Channel <b0cb59a63bf648fd@speechsynth> to Connection 172.16.1.179:1544 <-> 10.1.6.80:19413 [0] -> [1]

2021-10-13 16:44:03:812353 [DEBUG] Signal Message to [MRCP Server] [0x7fef240008c0;2;3]

2021-10-13 16:44:03:812382 [DEBUG] Wait for Messages [MRCPv2-Agent-1] timeout [600000]

2021-10-13 16:44:03:812403 [DEBUG] Process Message [MRCP Server] [0x7fef240008c0;2;3]

2021-10-13 16:44:03:812422 [DEBUG] Dispatch Signaling Message [1]

2021-10-13 16:44:03:812431 [INFO] Process SPEAK Request <b0cb59a63bf648fd@speechsynth> [1]

2021-10-13 16:44:03:812462 [DEBUG] Wait for Messages [MRCP Server]

2021-10-13 16:44:03:812853 [INFO] Initiate HTTP request <b0cb59a63bf648fd> guid [7b39a24c81944fddbc96a87857ab8b10] [https://uksouth.tts.speech.microsoft.com/cognitiveservices/v1]

<speak version='1.0' xml:lang='en-GB'><voice xml:lang='en-GB' gender='Male' name='Microsoft Server Speech Text to Speech Voice (en-GB, George, Apollo)'>I have of late wherefore I know not lost all my myrth twenty</voice></speak>

2021-10-13 16:44:03:813803 [INFO] Set HTTP request timer [30 sec] for <b0cb59a63bf648fd>

2021-10-13 16:44:03:980330 [INFO] Received HTTP response [403 Starting September 1st, 2021 standard voices will no longer be supported for new users. Please use neural voices for your speech synthesis request on cloud or on prem. To make the change please update the voice name in your speech synthesis request to the supported neural voice names in chosen languages.] in 167 msec for <b0cb59a63bf648fd> [https://uksouth.tts.speech.microsoft.com/cognitiveservices/v1]

2021-10-13 16:44:03:980387 [INFO] Server: openresty/1.15.8.2

2021-10-13 16:44:03:980395 [INFO] Date: Wed, 13 Oct 2021 15:51:17 GMT

2021-10-13 16:44:03:980402 [INFO] Transfer-Encoding: chunked

2021-10-13 16:44:03:980410 [INFO] Connection: close

2021-10-13 16:44:03:980430 [INFO] Strict-Transport-Security: max-age=15724800; includeSubDomains

2021-10-13 16:44:03:980439 [WARN] Request failed <b0cb59a63bf648fd>

2021-10-13 16:44:03:980446 [INFO] Clear HTTP request timer <b0cb59a63bf648fd>

2021-10-13 16:44:03:980456 [INFO] Close HTTP connection <b0cb59a63bf648fd>

2021-10-13 16:44:03:980603 [DEBUG] Signal Message to [MRCP Server] [0x7fef3c00dd00;3;4]

2021-10-13 16:44:03:980667 [DEBUG] Process Message [MRCP Server] [0x7fef3c00dd00;3;4]

2021-10-13 16:44:03:980690 [INFO] Process SPEAK Response <b0cb59a63bf648fd@speechsynth> [1]

2021-10-13 16:44:03:980703 [DEBUG] Signal Message to [MRCPv2-Agent-1] [0x7fef4402ef30;1;0]

2021-10-13 16:44:03:980734 [DEBUG] Wait for Messages [MRCP Server]

2021-10-13 16:44:03:980754 [DEBUG] Process Poller Wakeup [MRCPv2-Agent-1]

2021-10-13 16:44:03:980767 [DEBUG] Process Message [MRCPv2-Agent-1] [0x7fef4402ef30;1;0]

2021-10-13 16:44:03:980803 [INFO] Send MRCPv2 Data 172.16.1.179:1544 <-> 10.1.6.80:19413 [110 bytes]

MRCP/2.0 110 1 401 COMPLETE

Channel-Identifier: b0cb59a63bf648fd@speechsynth

Completion-Cause: 004 error

Ed James

unread,

Oct 13, 2021, 12:50:39 PM10/13/21

to UniMRCP

I tried changing the voice name to Wavenet (from https://cloud.google.com/text-to-speech/docs/voices ) but still got the same error:

<?xml version="1.0"?><speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xml:lang="en-GB"><voice name="en-GB-Wavenet-A">I have of late wherefore I know not lost all my myrth twenty</voice></speak>

Thanks

Ed

Ed James

unread,

Oct 13, 2021, 1:14:58 PM10/13/21

to UniMRCP

It looks like I'm talking to Azure and not Google. I’m not specifying a provider yet and I believe the server is supposed to use the first engine (which would be GSS-1) from the <plugin-factory> server config but maybe that’s not working:

<plugin-factory>

I’ll disable Azure just to double check.

Thanks

Ed

Ed James

unread,

Oct 13, 2021, 2:39:01 PM10/13/21

to UniMRCP

Hi All, yes that's what it was, if Azure is disabled then Google works.

Thanks

Ed

Arsen Chaloyan

unread,

Oct 26, 2021, 11:21:08 PM10/26/21

to UniMRCP

Reference to the other post in case anyone else hits the same problem with the order:

https://groups.google.com/g/unimrcp/c/m_vK5cc9T2k/m/We5VTZLUBwAJ

--
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+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/unimrcp/35abfe64-431d-40d9-86d3-6f31fd7bb64bn%40googlegroups.com.