Which package/tool you used for semantic search?
FAISS for nearest neighbor vector search.
Show HN: I trained an AI model on 120M+ songs from iTunes
241–250 of 444 posts
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#242Can it do endless mashups?
I think I can technically use the underlying model to do it, but never really tried it. Interesting...
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#243Wow great project. I tried it with a song and I kind of get what I expected. The results are more unique compared to other song search sites
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#244I tried one of my own songs... I suppose it picked up that my stuff is a bit jazz-influenced, but it otherwise doesn't seem very similar. It tickles me that it thinks I'm similar to both Johnny Mathis and Willie Nelson, though. :)
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#245This is FANTASTIC! There's a ton of naysayers here, but I'm going through songs and having a great time with this. It seems most forms of EDM work great with this setup! One funny thing is for heavily remixed tracks, all the remixes pop up as suggestions. :)
Thanks! The current model does tend to do well with EDM, but got a new model in the works that should hopefully address a lot of the shortcomings of the current one!
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#246I think I either found some kind of bug or Apple Music is really weird. The similar songs for Anaal Nathrakh - Endarkenment [0] have Annal Nathrakh with エンダーケンメント at the top. So a misspelling of the band, and apparently (according to google translate) a Japanese transliteration of the song title (Endākenmento, enderkenment). The song itself is literally the same one, others listed are their other songs, or more Japan…
Ya Apple Music can have some interesting duplicates in their catalog lol :/
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#247The search seems to be down? I'm not getting any results back in the dropdown.
Oof, honestly didn't expect the surge in traffic lol. It should work, as auto-scaling kicks in, etc. Sorry about that!
Ah just realized it doesn't work in Firefox. I think the autoscaling has kicked in, just results aren't appearing in Firefox.
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#248This is very interesting, but unfortunately I haven't had the greatest luck in finding new songs I would enjoy listening to. It absolutely finds similar sounding tracks, but it doesn't distinguish which part of the song made it enjoyable. There's no tempo consistency or genre consistency or even main instrument/vocal timbre consistency between recommendations. I think locking one or more of those dimensions would all…
Hey thanks for the feedback! I definitely have a lot of improvement to do on the model, it currently performs better for some styles/genres of music than others. But the model architecture I'm using is kinda outdated as well, gotta iterate on it more to improve it further! I'm also thinking of letting users upvote/downvote results, which can also help improve quality on the ranking side.
FWiW I had one shot and entered "Tabaran"
Rather than get back anything "acoustically similar" it simply returned a list of other songs on the same album (several of which are far from being acoustically similar).
No drama, you're attempting to cover a lot of ground, but I'm guessing there was no actual fingerprint there for that work and no sense of other songs that sounded similar.
ADDENDUM: Okay, I had to select the song .. but still "something went wrong" - perhaps hugged to death or not found to process. No matter :-)
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#249Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#250Wow great project. I tried it with a song and I kind of get what I expected. The results are more unique compared to other song search sites
Hey thanks for your feedback! :D