Suggestion : Add Genre Support about website HOT 6 OPEN

This seems pretty tricky. We have a back catalogue of over 1000 tracks that we'd also need to back fill, so we'd have to get genres from each of those before the index would be useful. @thelovebug and @ymauray - what do you guys think?

from website.

I concur on the trickiness. Whilst it might be possible to back-fill for the Jamendo tracks, considering that we have the source URLs, there are often multiple genres assigned to a single track, and I'm pretty sure that they're free-type, so would be difficult to filter by.

It needs thinking about.

from website.

This seems pretty tricky. We have a back catalogue of over 1000 tracks that we'd also need to back fill, so we'd have to get genres from each of those before the index would be useful. @thelovebug and @ymauray - what do you guys think?

One thing i thought might be possible would be to have some background process that pulls the ID tags out of either a jamendo page, or out of the audio file itself - For example, the BradSucks track Bad Attraction has a tag of #rock https://www.jamendo.com/track/30061/bad-attraction - which could be either scraped from the site, or use some provider (based on the source URL) to get the tags.

If that isn't possible, then looking in the MP3 might be possible :

using the site https://tagmp3.net/ and uploading the MP3 of bad Attraction reveals the value 255 - which means that Brad hasn't tagged this track properly. There is a PHP library for reading ID tags from files so that might be something.

managed to get an API key for Jamendo - was able to extract this data about the track :

from website.

We use the library for other tools, so we might as well give that go. Thanks for the suggestion.
As you have discovered, Jamedo files are not always - scractch that, almost never - properly tagged when it comes to genre. Also, genres on Jamendo are free form tags, and people tend to put whatever is popular and not really what it is - hence a lot of #rock tracks that are not rock at all :-)
So it's a case of deceiding if something that might not be accurate is better than nothing.
Again, I don't think it fits with the quality we'd like to have ton CCHits, but it is certainly something we might look into.

from website.

@all-contributors please add @computamike for ideas, usertesting

from website.

@JonTheNiceGuy

I've put up a pull request to add @computamike! 🎉

from website.