3 ms·

I've mostly come to the conclusion that basing music recommendations on bands/albums/songs just doesn't work for whatever my taste is. It seems like the priors

by thom 7y ago

I've mostly come to the conclusion that basing music recommendations on bands/albums/songs just doesn't work for whatever my taste is. It seems like the priors are so swamped by popular stuff that I rarely hear anything even approaching the songs I like. Apple Music's For You is particularly bad. Pretty sure in a moment of weakness I must have liked a Mumford and Sons song years ago because it thinks I love folk music. It wants me to listen to country. Data quality is clearly an issue because it thinks I like the rappers Frank Black and Gene because I like Pixies front-man Frank Black and 90s UK indie posers Gene. I am sure other services exist based on key, chords, dynamics etc - any recommendations?

jacobobryant 7y ago
I'm not sure off the top of my head, but that's definitely an active area of music recommendation research. There was someone in the lab at my university (BYU) who was/is working on a project he called pop*, although I'm not finding it with a quick google search. It's a system for composing new music, but part of the idea is that by figuring out how to make a system that understands the elements of music composition, it'll help in other areas like music recommendation too. As time goes on I am planning to integrate things like this into Lagukan. I think a major problem with mainstream recommenders is that they mix together the two problems of 1) deciding when and how often to play songs the user already knows, 2) recommending entirely new songs. Lagukan is an ensemble recommender, i.e. it combines different algorithms. The core algorithm I've made is focused entirely on problem #1, but the plan is to incorporate lots of different approaches (e.g. algorithms based on compositional elements) for problem #2, and then have Lagukan be able to adapt to what works for individual people.
chewz 7y ago
Exactly. It is also a question of a moment and granularity. I am in the mood to listen to certain kind jazz and I might be in a mood for a different music tomorrow morning. Most recommendations are either too broad (Louis Amstrong and Tony Bennett is not the same kind of jazz as Miles Davis or Weather Report) or too conservative (Youtube Music Your Mixtape is basically repeating same songs over and over never venturing out). And some seem more like paid adverts then recommendations. So please grow and give me good recommendations. And I will pay for it.
- jacobobryant 7y ago
  I'll do my best! If you haven't already, you can create an account at https://lagukan.com/docs/ https://lagukan.com/docs/ and then I'll send you updates as the project progresses.

by thom 7y ago

jacobobryant 7y ago

I'm not sure off the top of my head, but that's definitely an active area of music recommendation research. There was someone in the lab at my university (BYU) who was/is working on a project he called pop*, although I'm not finding it with a quick google search. It's a system for composing new music, but part of the idea is that by figuring out how to make a system that understands the elements of music composition, it'll help in other areas like music recommendation too. As time goes on I am planning to integrate things like this into Lagukan. I think a major problem with mainstream recommenders is that they mix together the two problems of 1) deciding when and how often to play songs the user already knows, 2) recommending entirely new songs. Lagukan is an ensemble recommender, i.e. it combines different algorithms. The core algorithm I've made is focused entirely on problem #1, but the plan is to incorporate lots of different approaches (e.g. algorithms based on compositional elements) for problem #2, and then have Lagukan be able to adapt to what works for individual people.

chewz 7y ago

Exactly. It is also a question of a moment and granularity. I am in the mood to listen to certain kind jazz and I might be in a mood for a different music tomorrow morning. Most recommendations are either too broad (Louis Amstrong and Tony Bennett is not the same kind of jazz as Miles Davis or Weather Report) or too conservative (Youtube Music Your Mixtape is basically repeating same songs over and over never venturing out). And some seem more like paid adverts then recommendations. So please grow and give me good recommendations. And I will pay for it.

jacobobryant 7y ago
I'll do my best! If you haven't already, you can create an account at https://lagukan.com/docs/ https://lagukan.com/docs/ and then I'll send you updates as the project progresses.