Abstract: A method includes receiving media content items. Using machine learning, audio segments are identified in the media content items based on analysis of content included in a corresponding media content item. Each of the identified audio segments is associated with automatically determined tags. A video clip is generated for a specific user using machine learning to automatically select, for the specific user, recommended audio segments from the identified audio segments based at least in part on prior user interactions of the specific user and the automatically determined tags of the identified audio segments, and identifying, for inclusion in the video clip, video segments of the media content items that correspond to the recommended audio segments. The video clip is provided for playback on a device associated with the specific user.

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: Media content episodes are received. Using machine learning, one or more media segments of interest are identified in each of the media content episodes based at least in part on an analysis of content included in a corresponding audio content episode. Each of the identified media segments is associated with one or more automatically determined tags. Using machine learning, a recommended media segment is selected for a specific user from the identified media segments based at least in part on attributes of the specific user and the automatically determined tags of the identified media segments. The recommended media segment is automatically provided in an media segment feed.

Abstract: The various implementations described herein include methods and devices for searching of audio content. In one aspect, a method includes obtaining a query string for audio content and obtaining a plurality of audio content results corresponding to the query string. The method further includes selecting a subset of results from the plurality of audio content results, including selecting respective search results from a plurality of sub-topic clusters, and sequencing the subset of audio content results. The method also includes causing the sequenced subset of audio content results to be presented to a user.

Abstract: The various implementations described herein include methods and devices for facilitating semantic search. In one aspect, a method includes obtaining audio content and extracting vocabulary terms from the audio content. The method further includes generating, using a transformer model, a vocabulary embedding from the vocabulary terms, and generating one or more topic embeddings from the audio content and the vocabulary embeddings. The method also includes generating a topic embedding index for the audio content based on the one or more topic embeddings, and storing the embedding index for use with a search engine system.

Abstract: The various implementations described herein include methods and devices for media segmentation. In one aspect, a method includes obtaining audio content for a podcast and generating sentence embeddings for the audio content. The method also includes generating segment embeddings using the sentence embeddings and context information, and determining, for each segment embedding, whether the segment embedding includes a topic transition for the podcast. The method further includes generating one or more topic transition timestamps for the podcast in accordance with the determining.

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: Media content episodes are received. Using machine learning, one or more media segments of interest are identified in each of the media content episodes based at least in part on an analysis of content included in a corresponding audio content episode. Each of the identified media segments is associated with one or more automatically determined tags. Using machine learning, a recommended media segment is selected for a specific user from the identified media segments based at least in part on attributes of the specific user and the automatically determined tags of the identified media segments. The recommended media segment is automatically provided in an media segment feed.

Abstract: Audio content episodes are received. Using machine learning, one or more audio segments of interest are identified in each of the audio content episodes based at least in part on an analysis of content included in a corresponding audio content episode. Each of the identified audio segments is associated with one or more automatically determined tags. Using machine learning, a recommended audio segment is selected for a specific user from the identified audio segments based at least in part on attributes of the specific user and the automatically determined tags of the identified audio segments. The recommended audio segment are automatically provided in an audio segment feed.