US Patent Issued to Google on March 11 for "Unsupervised parallel tacotron non-autoregressive and controllable text-to-speech" (American, Japanese Inventors)

ALEXANDRIA, Va., March 12 -- United States Patent no. 12,249,315, issued on March 11, was assigned to Google LLC (Mountain View, Calif.).

"Unsupervised parallel tacotron non-autoregressive and controllable text-to-speech" was invented by Isaac Elias (Mountain View, Calif.), Byungha Chun (Tokyo), Jonathan Shen (Mountain View, Calif.), Ye Jia (Mountain View, Calif.), Yu Zhang (Mountain View, Calif.) and Yonghui Wu (Fremont, Calif.).

According to the abstract* released by the U.S. Patent & Trademark Office: "A method for training a non-autoregressive TTS model includes obtaining a sequence representation of an encoded text sequence concatenated with a variational embedding. The method also includes using a duration model network to predict a...