Genesys Cloud - Developer Community!

Hello, 

We want to use Microsoft Azure Speech Services within Genesys, for TTS.

There is a standard connector (Microsoft Azure Cognitive Services text-to-speech (TTS) integration) but we can't use it, as it doesn't meet our internal minimal security requirements (the authentication is done through a simple subscription key). We are therefore looking at the recently released Genesys TTS connector that supports more authentication methods.

I did the setup but I'm getting an error: Failed to validate genesys-tts-connector integration. I asked the support who just told that they see "Unknown voice type" on their end, but nothing more (no additional documentation nor details). They told me to post here.

Did anyone managed to do such integration, or has any other example/documentation?

Thank you!

My current setup is:

• Synthesize Text URI: https://eastus.tts.speech.microsoft.com/cognitiveservices/v1
• List Voices URI: https://eastus.tts.speech.microsoft.com/cognitiveservices/voices/list
• Synthesize Text Request Method: post
• Synthesize Text Attribute: text
• Synthesize Voice Attribute: voice
• Synthesize Language Attribute: language
• Synthesize OutputFormat Attribute: format
• Synthesize Sample Rate Attribute: samplerate
• Voices Attribute: ./**
• Voice Name Attribute: ShortName
• Voice Id Attribute: ShortName
• Voice Language Attribute: Locale
• Voice Country Attribute:
• Voice Gender Attribute: Gender
• Voice Type Attribute: VoiceType
• PCI Compliant: False (default)

Hi Bertrand,
We will be doing this shortly, thus not able to provide any update yet, though I have a question. Do you need to provide values for the attributes, like "Voice Gender Attribute: Female", "Voice Country Attribute: US", "Voice Language Attribute: en-us", "Synthesize Language Attribute: en-us" etc.  ?

My understanding is that https://eastus.tts.speech.microsoft.com/cognitiveservices/voices/list is supposed to return all the voices and their attributes.

For example:

{
    "Name": "Microsoft Server Speech Text to Speech Voice (af-ZA, AdriNeural)",
    "DisplayName": "Adri",
    "LocalName": "Adri",
    "ShortName": "af-ZA-AdriNeural",
    "Gender": "Female",
    "Locale": "af-ZA",
    "LocaleName": "Afrikaans (South Africa)",
    "SampleRateHertz": "48000",
    "VoiceType": "Neural",
    "Status": "GA",
    "VoiceTag": {
      "ModelSeries": [
        "Monolingual"
      ],
      "PersonaId": [
        "Adri"
      ],
      "Source": [
        "Azure"
      ],
      "TailoredScenarios": [
        "Assistant",
        "Narration"
      ],
      "VoicePersonalities": [
        "Well-Rounded",
        "Animated",
        "Bright"
      ]
    },
    "WordsPerMinute": "147"
  }

And then, you need to map the values between Genesys and this. So for example, Voice Language Attribute: Locale so that Genesys knows where to look in the json for it. 

And for the Synthesize  one, it is how Genesys will build the POST call to Azure.

But it's only a guess!

Thanks. Would that mean if we need more than one language, we need multiple integrations, or expecting all language voices return in the single call.