Systems, methods and apparatus for generating music recommendations based on combining song and user influencers with channel rule characterizations
Systems, methods and apparatus for generating music recommendations based on combining song and user influencers with channel rule characterizations are presented. Such systems and methods output a playlist, which may be delivered as an information stream of audio on a user or client device, such as a telephone or smartphone, tablet, computer or MP3 player, or any consumer device with audio play capabilities. The playlist may comprise various individual audio clips of one genre or type, such as songs, or of multiple types, such as music, talk, sports and comedy. The individual audio clips may be ordered by a sequencer, which, using large amounts of data, generates both (i) user independent and (i) user dependent influencer weightings for each clip, and then combines all of such influencer weightings into a combined play weighting W for a given audio clip, for a given user. Taking the various play weightings W(Ui, Sj), a set of rules may be applied to generate a set of candidates C(Ui, Sj, Tk) to play to User j in each of Time slots k through k+m. Real time playlists may then be generated from the m sets of candidates by application of a set of rules, which may be channel rules, for example. The data used to generate influencer weightings may include user-specific data including preferences and detailed listening history, audio clip specific data, and data gleaned from various Internet accessible sources, including social media. In some embodiments a feedback loop may be implemented to gauge the accuracy of the dynamically generated playlists and modify the influencer weightings in response.
1. A computer-implemented method of generating an audio clip sequence to be played to a user, the method comprising:
selecting, by at least one processor, a set of audio clips;
calculating, by at least one processor, one or more user-independent weightings for each clip in the set;
calculating, by at least one processor, one or more user-dependent weightings for each clip in the set;
combining, by at least one processor, the user independent and user dependent weightings into an overall weighting for each clip, for a user, to generate an overall weighted set of audio clips;
selecting, by at least one processor, a set of candidate clips including a predetermined number of clips from the overall weighted set of audio clips, the selecting includes:
applying, by at least one processor, a set of rules having respective rule weights to the overall weighted set of audio clips and adjusting the overall weighting of each audio clip based at least in part on the applicable rule weights;
sequencing, by at least one processor, the selected set of candidate clips into a plurality of consecutive timeslots of audio clips to be played to the user;
selecting a next song based on a computed weight of a cluster of the candidate clips, wherein the next song is based on a highest computed weight; and
on a condition that at least one rule of the set of rules is violated by an audio clip, set the weight of that audio clip to zero, unless there are insufficient audio clips in the candidate set to generate an audio clip for a given timeslot.
2. The method of claim 1 , wherein the user independent weightings are based at least in part on at least one of:
time/basic dayparting, data obtained from social media/crowd/web scraping, song popularity distribution, revival perspective, resurgence of artist, societal events, artistic period, or aggregated internal global user profile statistics.
3. The method of claim 1 , wherein the user-dependent weightings are based at least in part on at least one of: user preference, user listening history, channel change, skip behavior, alerts/favorites/presets, user psychoanalysis, weather, version of music, mobile location, mood, upcoming events, user to user similarity, results of social media, crowd sourcing or web analytics analyses.
4. The method of claim 1 , further comprising at least one of:
offering to the user audio clips for play that have an affinity to a set of most recent media clips that were played,
offering to the user clips for play that have an affinity to a set of most recent media clips that were played, wherein the affinity is at least one of (i) channel to channel, (ii) channel to episode, (iii) episode to episode, (iv) artist to artist, (v) artist to song, or (vi) artist to channel, or
offering to the user clips for play that have an affinity to a set of most recent clips that were played, wherein the affinity is calculated based on frequency counts of said most recent songs played.
5. The method of claim 1 , wherein at least one of:
a tracklist is generated for each of two or more genres of audio content, and the tracklists are then combined in a mashup to generate a mixed audio content output;
or
a tracklist is generated for each of two or more genres of audio content, and the tracklists are then combined in a mashup to generate a mixed audio content output, and the genres of audio content include any of talk, music, comedy, sports, and news.
6. The method of claim 1 , wherein said selecting audio clips to obtain a candidate set includes one or more of:
(i) selecting some number of songs at random from the overall weighted set and applying the rules,
(ii) selecting all the songs in the overall weighted set,
(iii) selecting those songs in the overall weighted set above some defined threshold,
(iv) selecting the top 10% weighted songs in the overall weighted set, or
(v) selecting a defined number N from each quartile.