In this tutorial, you will learn how to set up a custom voice for text-to-speech (TTS) using Microsoft Azure Speech Services and Voice Gateway. The process involves training and configuring a custom voice.
Train your Custom Voice
Navigate to Microsoft Azure Speech Studio in the Azure Portal to train your custom voice using multiple utterances. After the training, deploy your custom voice, as you will require the deployment ID for the subsequent steps.
Configure your Custom Voice
Depending on your subscription, you can add the custom voice to your existing TTS Speech Vendor or create a new entry in the Voice Gateway Self-Service Portal. This choice depends on how your subscription is configured within the Microsoft Azure Portal.
Let's assume that the Custom Voice is configured as a separate resource from your default STT/TTS credentials.
To configure your custom voice, follow these steps:
- Log in to the Voice Gateway Self-Service Portal.
- Navigate to Speech > Add a Speech Service.
- Select Microsoft Azure Speech Services as your vendor.
- Define the accounts that should be given access to this vendor (the TTS service).
- Add a label to differentiate multiple API keys of the same vendor.
- Activate the Use for text-to-speech setting since it's necessary to utilize the Custom Voice feature, which depends on your subscription settings configured in the Microsoft Azure Portal.
- From the Region list, select the region of your subscription.
- Add your API key.
- Activate the Use a custom voice for TTS section and paste the deployment ID of your custom voice. You can find the deployment ID in your Custom Voice Endpoint URL within the Microsoft Azure Portal.
Now, you have two options on how to use your Custom Voice. You can use it as a default TTS Voice or configure it for specific outputs within your Flow.
Set up Custom Voice as the Default TTS
Within the VG Portal, go to the Application, which should use the Custom Voice as the default text-to-speech.
To set up your custom voice as the default TTS, follow these steps:
- Scroll down to the Speech Synthesis section and select your vendor. After selecting the label of the Speech vendor with custom voice enabled, you will notice that the Voice dropdown transforms into a text field.
- Enter your custom voice name, which you have configured within the Azure Portal.
Set up Custom Voice for Specific Parts of the Flow
To configure the custom voice for specific parts of the Flow, you can access Cognigy.AI within your Flow.
There are two options to change voices within your Flow:
- Use the Set Session Config Node.
- Adjust the Activity Parameters in the Say Node.
For both types, you must configure the Synthesizer settings and add your custom voice name.
Enable the advanced TTS Config toggle to set your deployment ID.
With this configuration set, every part of your Flow will use your custom voice as text-to-speech.