Match function working with Spotify but gives completely useless results when switched to iTunes.

can anyone from the team elaborate what the matching algorithms are based on?

Why do I get J.S. Bach piano music when I have a 80’s new wave song playing. Or why do I get a 90’s alternative rock track matched to a 10’s Hip hop track?

it works really nicely in Spotify but not when set to ‘iTunes’, let alone to ‘My Library’.

Please help me understand,

Thank you.

Hi Louis, 

The search results differ from the pool size which can be used in order to recommend other tracks.

As you can imagine, the results get better when you use the whole Spotify library compared to your personal iTunes library.

Using the iTunes library, things like analyzed Key, BPM, are being mainly used, that is why maybe this result is popping up.

Cheers,Lukas E.

Thank you for your feedback.

Forwarded this internally. We have a request for personalised Local Matching (matching preferences like Genre) where I added a vote.

Cheers,
Lukas E.

ah, I see. I have a rather large itunes library with 40.000 songs though and I am not expecting you to incorporate convolutional neural networks but would it be possible to include at least genre as well for the itunes matching? That way we could make sure to stay within the genre at least.

Thank you, Lucas! much appreciated!