Earlier quoted context omitted.
The preview audio is free. It may just be 120,000,000 previews.
Even so, how do you get the 120M previews?
... without getting banned?
Show HN: I trained an AI model on 120M+ songs from iTunes
331–340 of 444 posts
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#332How did you get the samples for the song? iTunes allows scraping? Your project is extremely motivational .. how long did it take you? What did you train on? I do DL for work and just play with things like cifar. This is so inspired.
most relevant
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#333[dead]
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#334I've never commented on HN before, but I feel compelled. Congrats, you get my first. I've been using your site for 10 minutes and already added song after song to my library. I'm sure you'll improve the matching algorithm over time. This is a great first step; I'm being exposed to songs I've never come across before. Good job!
Same for me, I don't comment often, but this is great! I like having instrumental music for when I'm working (with no vocals), so if the model could classify vocals vs no vocals that would make it even more useful for me.
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#335The idea is just great. Keep in mind what people expect from this type of website (and clear up any misconceptions), as this may lead to unnecessary churn. I would like to see some degree of customizability, e.g. weighting of features or something like that. (Nitpick: wrap the play svgs in a button tag like the "similar" button)
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#336Fantastic tool, thanks for making it!
As others said, it would be a nice option to export it as a playlist for Spotify/Apple/Youtube or a .txt file with "Artist - Track name" in each line. Then you can import the txt file into a playlist converter tool like www.tunemymusic.com and play it in your favorite music service. Mine is discoverquickly.com where you can listen Spotify playlists fast with mouseover, and discover related songs/artists.
An autoplay x seconds of every track, where you can choose the number of seconds, would be a nice addition too. This way you can discover music while doing other things.
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#337It seems to be finding snippets of songs similar to snippets. I tried on Metallica - Unforgiven III, which starts off with a slow piano composition, and then enters a riff and cuts out. It ends up recommending piano songs, many Korean ones. There's some interesting ones like Ghost - Cirice, where it finds other songs with similar riffs. I like Ghost's music in general, just not the Satanic themes, so this is a great…
good point, probably a final version should have a pipeline like 1. cluster song segments into styles and 2. search for each cluster or only the main cluster.
What would be a good NN architecture for the first step?
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#338Dream Theater Space Dye Vest
-> returns christmas music
yeah I think your model missed the mark lol
It needs to go off the bulk of the song, not the first 20 second preview of the song...
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#339Interesting that you linked Hans Zimmer, as he's known for "reusing" material across OSTs. A particular example that comes to mind (and where the AI did not find a match) is The Battle (5:52) from the Gladiator OST vs the Pirates of the Carribean OST.
Re: Show HN: I trained an AI model on 120M+ songs from iTunes
#340Earlier quoted context omitted.
Honestly it's loads better than current Spotify/YouTube Music suggestions. Mostly they just seem to suggest popular stuff that's heavily marketed...even though I seeded all my "thumbs up" with only eclectic stuff. Yes, it's hard to find a song I really really like, but 1-in-10 seem to be something I'd add to my eclectic "thumbs up" playlist. And almost none of them are by any artist that I've heard of before. This is…
You're not alone. For me, Spotify suggestions are "things you won't hate." Most everything is palatable, but forgettable and too usually not all that interesting.
I'd like to add, it's not all the platforms' fault. Too many artists aren't artists at all. The make too little effort to be unique.