| CPC G11B 27/031 (2013.01) [G06F 40/58 (2020.01); G06V 10/774 (2022.01); G06V 10/82 (2022.01); G06V 20/41 (2022.01); G10L 17/02 (2013.01); G10L 17/04 (2013.01); G10L 17/06 (2013.01); G10L 17/18 (2013.01); G10L 21/013 (2013.01); G10L 25/57 (2013.01); G10L 2021/0135 (2013.01)] | 20 Claims |

|
1. A method for generating training data for training a machine learning model to perform localization of audiovisual content, the method comprising:
receiving first content in a first language, the first content comprising audio and video data;
generating second language data by translating, using one or more processors, at least a portion of the first content to a second language;
generating, using a machine learning model, second content associated with a second presentation of the second language data;
evaluating the second content for one or more performance characteristics to generate a first performance score;
generating third content by adjusting a parameter of the second presentation of the second language data in response to evaluating the second content;
evaluating the third content for the one or more performance characteristics to generate a second performance score;
annotating the second content using the first performance score; and
annotating the third content using the second performance score.
|