Custom voices

Neural TTSaaS supports voices developed using the Microsoft Custom Neural Voice portal  .

Information required

After you create and deploy a custom voice, send the following information to the Microsoft Azure AI TTS Service team:

• VoiceName: The name of the custom voice created in Microsoft.

• EndpointID: The location of the deployed voice, returned by Azure as described in Deploy and use your voice model  .

• Gender: Male or Female.

• Language: The language code in two-letter notation, for example en-US for American English.

Optionally include these additional fields:

• SubscriptionKey (optional): A specific Azure subscription. If not specified, the default Azure subscription key is used.

• Visibility (optional): Whether the custom voice is considered public or private. The default is private. Public voices can be used by all users, but private voices are exclusively reserved to specific customers and require CCSS authentication.

An administrator will enter your custom voice into a Neural TTS configuration file and redeploy the service.

Synthesis requests

You can then use the voice in your synthesis requests by entering the voice name in SynthesisRequest > Voice > name.

Get voices requests

The GetVoices method returns information about public custom voices. For private custom voices, see your TTS service team.