Abstract: An electronic device obtains a listening history for a media item, the listening history including retention information indicating, for each respective portion of a plurality of portions of the media item, a number of listeners who listened to the respective portion of the media item. The electronic device, using the retention information, determines a pattern indicating a reduction in the number of listeners who listened to corresponding portions of the media item and determines a start time and an end time corresponding to a first portion of the corresponding portions of the media item. In accordance with the determination that the first portion of the corresponding portions of the media item meets predefined sub-content criteria, the electronic device stores an indication that the first portion of the media item comprises first sub-content, different from primary content, embedded in the media item.

Abstract: The various implementations described herein include methods and devices for speaker diarization. In one aspect, a method includes obtaining an audio recording and generating an embedding signal from the audio recording. The method further includes factoring the embedding signal to obtain a basis matrix and an activation matrix, including obtaining a sparse optimization of the embedding signal by minimizing a norm corresponding to the factored embedding signal. The method also includes generating a speaker log for the audio recording based on the sparse optimization of the embedding signal.

Abstract: A method includes retrieving a text from a database. The text corresponds to audio from a media content item that is provided by a media providing service, and the text includes a plurality of segments. The method also includes assigning a score for each segment in the text by applying the text to a trained computational model. The score corresponds to a predicted relevance of the respective segment to a narrative of the media content item. The method further includes identifying a non-narrative segment within the text using the assigned scores.

Abstract: An electronic device obtains a listening history for a media item, the listening history including retention information indicating, for each respective portion of a plurality of portions of the media item, a number of listeners who listened to the respective portion of the media item. The electronic device, using the retention information, determines a pattern indicating a reduction in the number of listeners who listened to corresponding portions of the media item and determines a start time and an end time corresponding to a first portion of the corresponding portions of the media item. In accordance with the determination that the first portion of the corresponding portions of the media item meets predefined sub-content criteria, the electronic device stores an indication that the first portion of the media item comprises first sub-content, different from primary content, embedded in the media item.

Abstract: An electronic device associated with a media-providing service obtains listening history for a media item. The listening history includes retention information that indicates a number of listeners who listened to the respective portion of the media item. Using the retention information, the electronic device generates a retention graph that represents the number of listeners who listened to corresponding portions of the media item as a function of time. The electronic device detects an extremum in the retention graph. The extremum corresponds to a reduction in the number of listeners who listened to the corresponding portions of the media item. The electronic device determines whether the extremum meets predefined criteria, and in accordance with the determination that the extremum meets the predefined criteria, the electronic device stores an indication that the portions of the media item corresponding to the extremum include sub-content embedded in the media item.