Yup, voice records are generated via Google speech API. Voice recognition is webkit speech input, only supported in Google Chrome. The application itself uses AngularJS.

This thing wouldn't have been possible without Google really.