Automatic selection of representative media clips

A clip selector, comprising:

• means for receiving an electronic media stream;
• means for identifying portions of the electronic media stream;
• means for determining human recognizable labels for the portions of the electronic media stream, where the means for determining the human recognizable labels includes:
• means for generating scores for the labels, where a score, generated for a particular label associated with a particular portion, of the portions of the electronic media stream, is indicative of a probability that the particular label is an actual label for the particular portion, and
• means for selecting the particular label as the actual label for the particular portion based on the score;
• means for gathering data, associated with user interaction with the electronic media stream, from a plurality of media players;
• means for determining at least one section of the electronic media stream with a particular media feature;
• and
• means for selecting a representative clip for the electronic media stream based on information regarding the labeled portions, the gathered data, and the at least one section.

A system, comprising:

• a clip selector to:
• receive an audio stream,
• identify a plurality of attributes associated with the audio stream,
• identify portions of the audio stream based on attributes associated with the audio stream,
• collect data regarding user interaction with regard to a plurality of audio players and the audio stream,
• identify at least one section of the audio stream that has a particular audio feature,
• generate scores for the identified portions of the audio stream, a score for the collected data regarding the user interaction, and at least one score for the identified at least one section of the audio stream that has the particular audio feature,
• combine the generated scores to obtain total scores for a plurality of sections of the audio stream,
• identify a particular section, of the plurality of sections of the audio stream, having a highest total score of the obtained total scores, and
• determine a representative clip for the audio stream based on the identified particular section, the determined representative clip including at least a portion of the identified particular section.

A computer-readable memory device containing instructions executable by a processor, the computer-readable memory device comprising:

• one or more instructions to receive an electronic media stream;
• one or more instructions to identify a plurality of attributes associated with the electronic media stream;
• one or more instructions to determine, based on the plurality of attributes and information regarding a plurality of previously-analyzed electronic media streams, for each of a plurality of instances in the electronic media stream, a probability that the instance is a beginning or an ending of a portion of the electronic media stream;
• one or more instructions to identify the portions of the electronic media stream based on the determined probabilities;
• one or more instructions to gather data associated with user interaction with the electronic media stream via a plurality of media players;
• one or more instructions to analyze features associated with the electronic media stream;
• one or more instructions to identify at least one section of the electronic media stream with a particular feature of the analyzed features;
• one or more instructions to generate scores for the identified portions of the electronic media stream, the gathered data associated with user interaction, and the identified at least one section of the electronic media stream with the particular feature;
• one or more instructions to combine the generated scores to obtain total scores for a plurality of sections of the electronic media stream;
• one or more instructions to identify a particular section, of the plurality of sections of the electronic media stream, having a highest total score of the total scores;
• and
• one or more instructions to select a representative clip for the electronic media stream based on the identified particular section and at least one of information regarding the identified portions of the electronic media stream, information regarding the data associated with user interaction with the electronic media stream, or information regarding the at least one section, where the selected representative clip includes at least a section of the identified particular section.

A method performed by one or more server devices, the method comprising:

• collecting, by one or more processors associated with the one or more server devices, information associated with user interaction with a video stream while playing the video stream via a plurality of video players;
• classifying, by one or more processors associated with the one or more server devices, the user interaction as at least two of positive user actions, negative user actions, or neutral user actions with regard to the video stream;
• generating scores for labels associated with portions of the video stream, where a score, for a particular label associated with a particular one of the portions of the video stream, is indicative of a probability that the particular label is an actual label for the particular one of the portions;
• and
• identifying, by one or more processors associated with the one or more server devices, a representative clip for the video stream based on the particular label and the at least two of positive user actions, negative user actions, or neutral user actions with regard to the video stream.