You can register your own voice AI model API and compare it with our provided APIs to evaluate model performance using the wizard.

Your API must return response data in one of the following formats:

  • Binary data (audio file)

  • JSON with base64-encoded audio file data stored in the audioContent field

    Example:

    {
  "audioContent": "//NEXAASSL5UAHpMTAJdcPZ6MCsePBW3MLg..."
}
1

Click "Register new API"


2

Enter your API information

Enter your API information (Provider name, Model name, API endpoint, HTTP method, HTTP headers, HTTP body)

For the HTTP body, you must include the template string {{SCRIPT}} in the script input field.

3

Click the Test button

Click the Test button to verify that your API is functioning correctly and returning the expected response.

4

Click "Add voice persona"

Click Add voice persona to register the voices and their information that your model supports.

5

Fill in the voice information

Fill in the voice information (Voice ID, Gender, Age, Speech style, Additional tags)

6

Click the Test button again

Click the Test button again to confirm the API is functioning, then click Save to store your configuration.

7

Click "See saved voices"

Click See saved voices to review all the voices you’ve saved so far.

8

Click Next and choose model

Click Next to navigate to the Choose model page, where you can select your newly registered API.

That’s it! You can now create evaluations using your own API.