Create a custom keyword using Speech Studio
Your device is always listening for a keyword (or phrase). For example, "Hey Cortana" is a keyword for the Cortana assistant. When the user says the keyword, the device sends all subsequent audio to the cloud, until the user stops speaking. Customizing your keyword is an effective way to differentiate your device and strengthen your branding.
In this article, you learn how to create a custom keyword for your device.
Create your keyword
At the Custom Keyword page, create a New project.
Enter a Name, an optional Description, and select the language. You will need one project per language and support is currently limited to the en-US language.
Select your project from the list.
To start a new keyword model click Train model.
Enter a Name for the keyword model, and optional Description and type in the Keyword of your choice, and click Next. We have some guidelines to help choose an effective keyword.
The portal will now create candidate pronunciations for your keyword. Listen to each candidate by clicking the play buttons and remove the checks next to any pronunciations that are incorrect. Once only good pronunciations are checked, click Train to begin generating the keyword.
It may take up to thirty minutes for the model to be generated. The keyword list will change from Processing to Succeeded when the model is complete. You can then download the file.
Save the .zip file to your computer. You will need this file to deploy your custom keyword to your device.
Test your custom keyword with Speech Devices SDK Quickstart.