A method includes providing, by an audio playback interface, an initial playlist comprising audio tracks. The method includes receiving a user preference associated with an initial audio track during a listening session, wherein the user preference is indicative of a listening mood of a user and comprises one or more of a user behavior or a natural language input. The method includes generating a representation of the user preference in a joint audio- text embedding space by applying a two-tower model comprising an audio embedding network and a text embedding network. A proximity of two embeddings is indicative of semantic similarity. The method includes training a machine learning model to generate an updated playlist responsive to the listening mood of the user during the listening session. The method includes applying the machine learning model to generate the updated playlist. The method includes substituting the initial playlist with the updated playlist.
The invention provides a video music matching method and device, electronic equipment and a computer readable storage medium, and relates to the technical field of video processing. The method comprises the following steps: inputting a video to be subjected to music matching into a pre-trained first neural network model to obtain specific-dimension video features of the video to be subjected to music matching; inputting the video features of the specific dimension into a predetermined second neural network model to obtain an emotion category of the video to be subjected to music matching; obtaining a plurality of songs corresponding to the emotion categories from a song library pool, and respectively extracting audio features of the songs; and performing Euclidean distance calculation according to the specific dimension video features of the video to be subjected to music matching and the audio features of each track, and taking the track of which the Euclidean distance is within a preset range as a recommended track of the video to be subjected to music matching. According to the invention, based on the emotion category expressed by the video, automatic music matching can be carried out on the video, and the use experience of a user is improved.
The invention provides a video file processing method and system, a medium and electronic equipment. The method comprises the steps of obtaining voice comment information input by a user for a current video file, wherein the voice comment information comprises voice content, voice duration and comment mood; identifying the content of the current video file, and generating a plurality of video scenes; determining the video scene matched with the voice comment information; and when the video file is played to the video scene, outputting the voice content. According to the method, the interactive interestingness of the reviewer can be improved; and user viscosity can be further increased.
Provided in embodiments of the present disclosure are a video processing method and apparatus, a device, and a medium. The method includes: receiving a play trigger operation for a forwarded video, the forwarded video being forwarded from an original video by a forwarding user; obtaining the original video, and forwarding comment information provided by the forwarding user on the original video when forwarding the original video; and playing the original video on a play interface of video works of the forwarding user, and displaying the forwarding comment information. In the embodiments of the present disclosure, when the forwarding user forwards the original video, the original video can be played when receiving a play station for the forwarded video, and meanwhile, the forwarding comment information provided by the forwarding user on the original video during the forwarding is displayed while playing the original video, instead of only displaying the comment information of the creator as in the related art. Thus, in this case, it is convenient for the user to know the forwarding user's feeling about the forwarded video, thereby improving the user's ense of experience.
The invention provides a video recommendation method and device, electronic equipment and a computer readable storage medium, and the method comprises the steps: obtaining the sensory information of a user when the user is detected to watch a video; sending the sensory information to a server, receiving related information of the to-be-recommended video returned by the server, and displaying the related information of the to-be-recommended video to the user, wherein the to-be-recommended video is determined based on sensory information. Scheme of the present disclosure, when it is detected that the user watches the video, the sensory information of the user is obtained, the watching state of the user, including the watching feeling of the currently watched video and the to-be-recommended video determined based on the sensory information of the user, can be reflected on the basis of the sensory information, and the watching feeling of the user is considered, so that the recommended video is more accurate and more accordant with the mind of the user.