Custom Voice

Build a recognizable, one-of-a-kind voice for your text-to-speech apps, as unique as your business.

New to Speech service? Create a Speech resource

Create a brand voice that supports custom personas

To customize your voice agent, simply record audio and upload it as training data. We do the rest - creating a unique voice font tuned for your recording. You can also develop a highly realistic, humanlike custom voice by using Microsoft’s groundbreaking neural text-to-speech models.

Hear Custom Voice for yourself

Build a highly natural voice without a single line of code, starting from just a few minutes of audio.

NEURAL ^(PREVIEW)

1 hour of speech data

HIGH-QUALITY

8 hours of speech data

STANDARD

3 hours of speech data

BASIC

2 hours of speech data or less

Paul

Zira

Evan

Zira

Mark

Jessa

Guy

Custom Neural Voice capability is now in public preview – with limited access

These voices are so lifelike, that they must be designed in a way to earn the trust of others. To learn about the principles of building synthetic voices that create confidence in your company and services, as well as responsible deployment of these voices, visit Microsoft's Responsible AI guidelines.

Apply now

Develop a highly realistic custom voice for your business

Brand identity

Design and implement a voice model that strengthens your brand strategy.

Custom persona

Extract value from your analytics to work by accommodating customer sentiment with custom voice characteristics.

Natural interaction

Increase your customers’ emotional connection and interactions with your applications.

Creating a custom voice model

1 Prepare training data and create a Speech resource before you start to train a Custom Voice.

2 Upload your data to the Custom Voice portal or through the Custom Voice API and check quality.

3 Use your data to train a custom model. Test the model with your script when it’s ready.

4 Deploy the voice model to get your custom API endpoint. Test the endpoint before you integrate it in your system.

5 Use the voice in your apps by using code samples from the endpoint.