US 11,960,536 B2
Methods and systems for organizing music tracks
Mikael Henderson, Copenhagen K (DK); Peter Berg Steffensen, Copenhagen K (DK); David Stubbe Teglbjærg, Copenhagen K (DK); Jesper Steen Andersen, Copenhagen K (DK); Jose Luis Diez Antich, Copenhagen K (DK); and Thomas Jørgensen, Copenhagen K (DK)
Assigned to Moodagent A/S, Copenhagen K (DK)
Appl. No. 17/764,603
Filed by Moodagent A/S, Copenhagen K (DK)
PCT Filed Sep. 30, 2020, PCT No. PCT/EP2020/077377
§ 371(c)(1), (2) Date Mar. 29, 2022,
PCT Pub. No. WO2021/064026, PCT Pub. Date Apr. 8, 2021.
Claims priority of application No. 19200375 (EP), filed on Sep. 30, 2019.
Prior Publication US 2022/0350838 A1, Nov. 3, 2022
Int. Cl. G06F 17/00 (2019.01); G06F 16/638 (2019.01); G06F 16/68 (2019.01); G06F 16/683 (2019.01); G10L 15/22 (2006.01)
CPC G06F 16/639 (2019.01) [G06F 16/683 (2019.01); G06F 16/686 (2019.01); G10L 15/22 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method for organizing music tracks, the method comprising:
providing a client device, and a storage device in data connection with the client device;
providing, on the storage device, a plurality of music tracks, each music track having linked therewith a feature vector and metadata, wherein the feature vector comprises feature values representing semantic characteristics of the respective music track, and the metadata comprises at least one type of metadata record representing associated information about the respective music track;
providing a similarity matrix, the values of the similarity matrix representing a measure of similarity between different metadata records of the same type;
receiving a query from the client device comprising an input vector, and at least one input metadata record;
selecting an ordered set of music tracks from the storage device by determining a number nm of similar metadata records with respect to the input metadata record using the similarity matrix;
determining a qualified tracks pool comprising, for each similar metadata record, a number nq of qualified music tracks from the storage device which comprise the respective similar metadata record in their metadata;
determining an order of music tracks within the qualified tracks pool based at least in part on a distance between the whole or part of the feature vector of each qualified music track with respect to the input vector in the vector space; and
returning to the client device, as a reply to the query, a playlist based on the ordered set of music tracks.