Searching for covers of songs.

  • 3
  • Problem
  • Updated 2 months ago
Back in the day, we were able to search for cover versions of specific songs. How amazing was that!? I would never know that Cibo Matto does a cover of "The Candy Man" from Willy Wonka, otherwise. This needs to be reinstated.

Along with an "and" option for the advanced genre search. This can't be THAT hard. Thanks for even reading this...
Photo of Adam Wacholder

Adam Wacholder

  • 28 Posts
  • 0 Reply Likes

Posted 6 years ago

  • 3
Photo of Andy DeNardi

Andy DeNardi

  • 231 Posts
  • 20 Reply Likes
I believe that I already gave you these links.  Dedicated sites like these do a much better job than AllMusic ever did.  This forum is littered with complaints of incorrect credits and composers.
 http://www.secondhandsongs.com/
http://www.coversproject.com/

And I've seen your request for an advanced search AND option several times.  Either it IS that hard or management doesn't agree with you.  I know how it feels.  Frankly, the regular search sucks so bad that I'm not sure your suggestion would turn out the way you envision it.
Photo of Adam Wacholder

Adam Wacholder

  • 28 Posts
  • 0 Reply Likes
Ah yes you might have. I've been reorganizing my bookmarks so thank you! The genre thing is dearest to my heart, as you can probably tell. Just try to work it in everywhere at this point...saturate the market.

Anyway, as a thank you check out musicroamer.com It's a nifty related-artist graphic web interface. Best of its kind I've found, hopefully its new to ya!
(Edited)
Photo of Adam Wacholder

Adam Wacholder

  • 28 Posts
  • 0 Reply Likes
The core issue is that AllMusic's database is the largest and most complete. This makes their disjointed interface and struggling search system just so..... TRAGIC.
Photo of Andy DeNardi

Andy DeNardi

  • 231 Posts
  • 20 Reply Likes
I agree that it's the most complete.  It's also frightfully inaccurate and by licensing that database to hundreds of web sites, it spreads those inaccuracies all over the web to the point that you can't find any correct information.  Wikipedia sometimes is better but many there borrow from AllMusic/Rovi as well.

I blithely submitted corrections for many years, but few are acted upon and looking at this forum, one can see that there are many others who have become frustrated after sending data multiple times without result.

I use this database a lot.  At least ten times every day, sometimes thirty or forty.  But it is only one of my tools.  In the past month, the site has been unavailable at some point every single day.  Nearly every problem posted here is foisted off on someone else - Rovi, the ad provider, the server farm.  Things have gone seriously downhill.  I will say that I am pleased with the most recent redesign, especially compared to the previous one.  But they're putting a new coat of paint on a burning building.

I've bookmarked the link for MusicRoamer and I think it will be very helpful.  Thanks!
I've got the last century's music nailed down but the last twenty years has been hard for me to keep up with because I don't follow pop or hip-hop any more.
Photo of Andy DeNardi

Andy DeNardi

  • 231 Posts
  • 20 Reply Likes
Reference point for my statement that the database sucks:
  • Persia My Dear (1987) -  Richard Davis / Roland Hanna / Freddie Waits
no rating, no review, 1 release listed, 3 artists credited
  • Persia My Dear (1987) - Richard Davis Trio
3 and a half stars, reviewed, same release listed but different release date, 9 artists credited
  • Persia My Dear (1987) - Roland Hanna
4 and a half stars but no review, 1 release with different number, 1 artist credited

Of the 48 "similar albums" provided for each entry, I found one album that was common between two of the three.  Demonstrating that the feature produces vague recommedations with little real science or thought behind them, generally random entries within a broad genre.

And as a cite for how difficult they have made it to correct errors, they know longer show record IDs so you can't directly point out releases or albums that disagree.

Before Advanced Searches can be meaningful, some stringency needs to be applied to the existing data..  My example is on the high end of the error scale; the majority are spelling or credit errors.  But I wasn't saving this one.  It came up just today in response to a query.  AllMusic/Rovi relies on crowdsourcing to correct errors, but makes it too difficult to submit them, is slow to imp[lement them and there is no no feedback asking for clarification or providing encouragement to continue.  It's just a black hole. 

Seemingly the only thing that AllMusic does in-house is write blog entries, and everything exists solely to serve ads to us.  Granted, all of this is free, but there is a lack of commitment to the reason we came here - the music.  They're just waxing the Studebaker.





Photo of Adam Wacholder

Adam Wacholder

  • 28 Posts
  • 0 Reply Likes
That is the best damn closing line ever. A perfect metaphor for your unfortunate truthbomb.
Photo of Diana Gregory

Diana Gregory

  • 2 Posts
  • 0 Reply Likes
What is so weird to me is that it *used* to work.  It may have worked two years ago - but not now - at least not what I was looking for.  This one should be simple - "Darkness, Darkness" by Jesse Colin Young.  I can't get any cover versions by searching for the song; and I do know of at least 3.  Plus - all the ones by Jesse are listed as being issued in 1995 - not 1965. It's as bad as Pandora using LyricsFreak - or whatever - and getting the wrong song.  I have some understanding why the Event Finder can't distinguish between UFO and U.F.O. (at least I can stop looking now that Phil has retired), good database searches strip all punctuation and force all upper or all lower.  But back when - if I were to look for the song and click on it (like the "Happy" mentioned), I'd have gotten a list of *all* song titles HAPPY - and the composer (and I think from there - maybe a sort on the list so you could have all the Jesse together, and all the Youngbloods together, and ignore anyone else, so you would be able to find covers of the song you are looking for - without having random band names or album names showing up.
Photo of Zac Johnson

Zac Johnson, Official Rep

  • 3378 Posts
  • 195 Reply Likes
Hi Diana.
If you go to the song page for "Darkness, Darkness" by The Youngbloods (Jesse Colin Young's band who originally recorded the song) you will see a listing for the song released in 1969. If you click on the "Also Performed By" tab, you will see a listing of other artists who have recorded this song.

https://www.allmusic.com/song/darkness-darkness-mt0028420256/also-performed-by




Jesse Colin Young re-recorded a version of the song as a solo artist in 1995 on the album Crazy Boy. You can see the listing for his performance of the song and other artists who covered it here:

https://www.allmusic.com/song/darkness-darkness-mt0013244990/also-performed-by




Photo of Zac Johnson

Zac Johnson, Official Rep

  • 3384 Posts
  • 196 Reply Likes
Hey Adam,
Ask and four years later ye shall receive.

We've been putting some back-end effort into generating a database of cover versions of songs. Some detail is here:
https://getsatisfaction.com/allmedianetwork/topics/improved-song-pages-on-allmusic

And here's the Candy Man page (complete with Cibo Matto cover): https://www.allmusic.com/song/the-candy-man-mt0007882774/also-performed-by
Photo of Adam Wacholder

Adam Wacholder

  • 28 Posts
  • 0 Reply Likes
Wow, you guys are wild! 4 years later and it works. Sort of. Thank you, Zac and friends.

I just tested it:
This page contains the oddball cover of The Beat Goes On by french band Casino Music. I was worried it was too rare to appear. https://www.allmusic.com/song/the-beat-goes-on-mt0010183020/also-performed-by

But searching for the Jolly Boys cover of Golden Brown (The Stranglers) I had to go to artist page directly to find it on its own. https://www.allmusic.com/song/golden-brown-mt0038767335

I guess depending on how certain song credits were standardized will affect how/if they appear. (By the way, if you guys are hiring I'd love to help clean up your database.)

Last one and I'll stop nitpicking, found a similar issue with Brian Eno's "By This River." The original album song link only shows one cover artist (https://www.allmusic.com/song/by-this-river-mt0008653725/also-performed-by), while the Yoshida Brothers version I was looking for shows 4 (https://www.allmusic.com/song/by-this-river-mt0029880475/also-performed-by).

I know, I'm so ungrateful!! But seriously, thank you. It still works how it should with the database information as it is. And I would seriously love to work with/for you to fix this if I could. Consider this my cover letter.
Photo of Zac Johnson

Zac Johnson, Official Rep

  • 3384 Posts
  • 196 Reply Likes
You've hit the nail on the head, Adam. Our logic takes as many connections and infers as much as we can in order to try to combine different line items in the database to group them into the same "song" but it is reliant on the core metadata available to us.

Just as "Happy" composed by Mick Jagger / Keith Richards of the Rolling Stones and "Happy" composed by Pharrell are different compositions by different performers, our data reflects that we have one "song" called "Golden Brown" composed by The Stranglers and one "song" called "Golden Brown" composed by H. Cornwell / D. Greenfield.
 
There is almost certainly a good reason for this. I bet on some (all?) Stranglers albums, the song is credited as being performed by the band (or maybe on the liner notes it says "all songs written by The Stranglers"), whereas on the Jolly Boys album the liner notes say H. Cornwell / D. Greenfield. Music fans may be able to listen to the two songs and know that they are the same, but we're reliant on the metadata associated with the tracks to determine the uniqueness of the song/composition.

The tricky part comes with resolving these. Should we encourage our data provider to honor the liner notes as they were published? Or do we ask them to overwrite the historical printed data to better reflect the true song entity (the words and melody as written by specific musicians). Or is there a third option where we can associate different performances with our internal SONG_ID and override the key that is automatically generated by the Track Title + Composers metadata?

And then we have to figure out how this applies to over 30 million tracks.

So as you can see, it is a good challenge. We'll keep working to refine the process and boil these individual instances down to try to provide the best solution.
Photo of Adam Wacholder

Adam Wacholder

  • 28 Posts
  • 0 Reply Likes
Thank you for this wonderful breakdown. Now serious question, how would I join the "team" working on this and other metadata issues? I am currently working in medical inventory, but would much rather be doing this. Seriously.
Photo of Zac Johnson

Zac Johnson, Official Rep

  • 3384 Posts
  • 196 Reply Likes
We license all of our music metadata from a company called TiVo (formerly Rovi, formerly Macrovision, formerly All Media Guide). They're the ones who manage all of the corrections and cleanups. Their Jobs page is located here: https://www.tivo.com/jobs

It doesn't look like they have any Technical Editor positions open now, but to tell you the truth, while this kind of in-the-weeds cleanup is certainly valuable to music fans and researchers, the company as a whole doesn't spend a ton of resources on this level of hair-splitting correction.

Even though it might drive guys like you and me crazy to see these small things that could be patched up, there isn't a real business driver to put a big team together to comb through the data at this level. 

From the website side we plan to keep refining the process and trying to leapfrog any data inconsistencies we can.