Virtuoso authorblues
Generate playlists based on related songs.
Uses Last.fm data to generate playlists based on similar songs and a naive matching algorithm. Performance improves greatly if your tracks are tagged accurately (Track name and Artist name are the only things required for Virtuoso to work, and they should match Last.fm's format).
Using Virtuoso is simple!
The following improvements will be implemented in future versions:
Please, rate this add-on after you use it, then leave a comment telling what you like and dislike, to help with future improvements. If there are any bugs, pass them along.
@NK4551, updated the description. the important thing is really the track and artist names. if they are spelled differently that Last.fm knows them, or Last.fm doesn't have information on those artists, then this will not work. i will work on getting it to match those titles a bit less rigidly.
I got enough songs to crash most of songbirds plugins which have to process all of them, but this plugin seems to be not able to build a playlist from it. I always tagged all songs really carefully, so I am sure that 99,9% of the songs got a perfect artist and title tag.
I had to run virtuoso about 20 times with different songs... at the ~20th time it happend the first time that virtuoso found more than 8 matching songs and was able to build a 10 songs playlist.
This playlist it generated consist of 7 songs from the same album and 3 songs of another album from the same artist.
This result is much worse than just listening to a single album ...
@inoX
well, unfortunately, the only data source that this plugin draws from is Last.fm's similar track data. i am working with their similar artist data as well, trying to get a good balance for tracks that dont have a match in the Last.fm databases.
really the only possibilities are that your tracks are obscure (i have a hard time generating playlists from some of my indie rock music, just because there is very little data for those artists in Last.fm) or that they are tagged slightly different from Last.fm (i had this problem with belle and sebastian: they were "belle & sebastian" in my library, and "belle and sebastian" in Last.fm)
the problem would solve itself if Last.fm attempted to use their own spelling correction algorithms to fix API input, but they dont. i have been experimenting with using google's json real-time spelling correction algorithm for use in their search box. this might help with some of the small spelling issues, like letter deletion, insertion, or transposition, as well as substitution of things like "&" or "+" for "and".
tl;dr version: im working on it, but it proves to be difficult without the help of Last.fm's spelling correction algorithm that they offer on their web interface.
Most of the results i got was permutations of the album, likewise inoX.
Nevertheless it's an interesting Add-On, please keep on developing :-)
@Antiheld
do you have any music by similar artists to inoX. if you could tell me some music you have that you thought should have appeared in that list but didnt, it might help me make improvements. thanks! :D
It works really well, though sometimes it seems to notice a song is related to another, but not vice versa, for what ever reason. Also, the maximum playlist size also seems to be the minimum playlist size. I suggest adding a separate option for minimum size and the ability to increase the maximum size beyond 50 songs.
Excelent idea and nice outcome. You could evolve it to a great addon! Congratulations
No matter which track I try, I get zero results The error is:
Searching for tracks similar to "Reason to Live" by Blackhouse
>>Found 0 track(s) by searching for "Reason to Live".
>>Added 0 new/unique tracks to the playlist.
No playlist could be generated from "Reason to Live".
No matter what track I choose I get a similar error. These tracks all worked fine with "Similar Artists" (that was the previous plug-ins name wasn't it?)
@nexist: i am not the creator of any other plug-ins. this tool ONLY uses Last.fm's similar tracks tool, so you have to pick tracks with at least a little notoriety.
@dakial: thanks for the kind words. more people comment when there is something wrong then when it works perfectly, so its nice to know that people enjoy my hard work.
@jeslimak: i really ended up using that number as the target size. i should change that (eventually)
Oh, the description for the other plug in says that this is the new version of that older plug-in (in the comments section). Well then, I guess this tool won't work for me then, since I am not going to change my musical tastes. While I do listen to some pop music I usually know what Queen or Scorpions album I want to hear.
based on 10 ratings
NK4551
It seems really cool, but I have such a small library that it can't form any playlists from my music (it also may be due to the fact that none of my metadata has a genre).
I think it would be nice if you could post what you mean by "tracks tagged accurately."
but it's super promising, and I won't get rid of this add-on any time soon.