Forum » Feedback and Ideas

track length relevancy

 
  • track length relevancy

    I apologize if this has already been suggested, but the search was not functioning.

    For the average song, around the standard 3-4 minutes, the current system works great. I feel though that listening to bands of varied song lengths can overpower something that has a LOT of play. Two bands which I listen to quite often, probably about the same amount of time, one has songs around 30-90 seconds where the other has songs around 15-25 minutes. One of them is ranked at #1 and I don't even know if the other has even made it on to my list, making lists very innaccurate.

    When the plugins send the tag info, if it also sent the track length, audioscrobbler could keep a running total of time for each artist, making that the primary method of ranking rather than total number of plays.

    Implementing this would void prior data, so something like this could be done: Because the system is based upon individual tracks and for each user in the database there is a list of every track they have played, as soon as other users play that track, it establishes a length.

    It'd be great seeing 15 days 23 hours 31 minutes of play along side 9,384 total plays!

    • DoctorT sa...
    • Användare
    • 6 sep 2005, 01:45
    You are about the gadzillionth person to make this suggestion. The developers have said nice idea, they'd like to do it, but they can't do it now.


    Music, like language, is universal.
    Redigerad av DoctorT den 6 sep 2005, 01:46
  • Sweet. Now when the search starts working again you can delete my post.

    -Gadzillion first

    • DoctorT sa...
    • Användare
    • 6 sep 2005, 02:20
    I know its frustrating with the search out, I've had problems finding things myself. Search is one thing I know they are working on and they will bring it live when they think it's ready.


    Music, like language, is universal.
  • Good deal. It seems like messageboard searches are always the first thing to stop working. Heh, I don't know.

    In my opinion, repeated threads, as much as people hate them, can serve a fairly decent purpose. If there is something the masses want, it will be noticed.

    • AndyChrz sa...
    • Användare
    • 6 sep 2005, 05:17
    This is one of the few rejected arguments that i have always supported, even though Russ has repeatedly said "Audioscrobbler/Last.fm is NOT a competition."

    Truth, but that doesn't mean that it won't make things look more accurate. As much as i like Pig Destroyer (i do), i honestly listen to Isis just as much if not more... but since PD has songs around 60-90 seconds long and Isis has songs around 6-9 minutes long... you wouldn't really know it.

    I dunno, this idea will get shut down literally hundreds of times. But i, for one, think that threads like this should continue to come up just to show that many many users are in favour of it.

    I remember on the old AS stie when they showed pics of the really really old site... and they used to do this. They used to have playing time on there (somehow, i don't quite remember exactly). And it was pretty cool.

    Eh, this'll get nowhere... but it's still a decent argument that i think quite a few support.


    Here are the awesome record company groups that i lead here on Last.fm... Black Market Activities, Eulogy, Ferret, Ironclad, Lambgoat, Mediaskare, Metal Blade, Prosthetic, Roadrunner, Sumerian, Thorp, Trustkill, Willowtip
    • Oldbean sa...
    • Användare
    • 6 sep 2005, 07:55
    How about just an average time modifier for every artist.

    • Feanor sa...
    • Moderator
    • 6 sep 2005, 08:01
    If the admins added in the track time, then people would just start complaining about other things. Track length isn't the only thing that affects your play counts, the amount of music the artist has released affects it too. I like Iron Maiden and Bruce Dickinson's solo stuff equally, yet BD is at #20 on my chart, while Maiden takes the top spot because they have released far more stuff. Do you expect them to take things like that into account as well?

    If you're a Kiwi, please join the Hurrah for Kiwialism group | Groups for cool people:this this this this
    • Russ sa...
    • Alumni
    • 6 sep 2005, 09:07
    I'm not bloody rejecting it, for the last time!

    What I'm saying is that it's impractical to do at this time. It'll take a lot of effort for very little gain. But we are considering it, and I'd like to do it at some point, but there are more important things to do.

    We also have no convenient way of getting the track length. Plugins do actually submit it, but we've got no mechanism to aggregate that data sensibly at the moment (we just store the first ever track length submitted). MusicBrainz has very few track lengths in their database.

    • [Raderad användare] sa...
    • Användare
    • 7 nov 2005, 15:24

    This is a subscription critical issue for me

    I know there are probably loads of stuff on your todo list, but having my profile reflect my taste is the most critical issue for me.

    I want my profile (and my personal radio and neighbours) to reflect what I'm listening to the most without having to manipulate the submissions.

    This is obviously "very little gain" for most people, but for fans of certain styles of music this is a big issue.

    • Russ sa...
    • Alumni
    • 7 nov 2005, 15:36

    Re: This is a subscription critical issue for me

    Quoth vesal:
    I know there are probably loads of stuff on your todo list, but having my profile reflect my taste is the most critical issue for me.

    I want my profile (and my personal radio and neighbours) to reflect what I'm listening to the most without having to manipulate the submissions.

    This is obviously "very little gain" for most people, but for fans of certain styles of music this is a big issue.


    OK, we appreciate it, but one person's subscription money isn't really going to change it for us :). We're still considering it though.

    • ahoier sa...
    • Användare
    • 7 nov 2005, 16:25
    with online music, and ripping, i think the times will always be different :P

    • karit sa...
    • Användare
    • 7 nov 2005, 20:47
    Well most prob a good start will be for everyone to check musicbrainz for their favorite artists and check that track lengths are there. And then tag all their music with the musicbrainz ids.

    But then again the servers seem to be working pretty hard at the moment so not sure if it is nice to give the servers even more work to do.

    • [Raderad användare] sa...
    • Användare
    • 1 dec 2005, 05:02
    Not to harp on an old idea, but looking at just the music I have on my work computer:

    Ludwig van Beethoven: 9 tracks, 5:47:25
    Dead Kennedys: 113 tracks, 5:25:05

    By my current charts (minus those artists), if I played these files Dead Kennedys would have enough plays to get #1, and Beethoven wouldn't make the top 50, despite the fact I would have listened to 23 more minutes of the Beethoven!

    This is a worst case example, but there are heaps of artists where the average song length would double that of other artists. Given the whole point of the site is to track the most played artists, I'd have thought this would be the #1 priority to fix.

    It shouldn't even be that hard. Make track time a required field in the 1.2 plugin protocol (even better, make it a requirement to track total played time. If a user seeks a song it should still record the amount of time they actually listened to it!) and add a database field to capture it on every submission. For existing data and submissions on the 1.1 protocol missing it, populate with the existing data from the first time submitted. It'll be slightly inaccurate, but much better than just recording tracks and not time.

  • I don't see this being any more accurate than counting by total tracks played per artist:

    Say I have a one hour track by band A, if I start listening to it and decide after 5-10 minutes I don't like it and press skip, the plugin will have already submitted a one hour play of artist A seen as we've crossed the 4min threshhold. Now band B have 5 min long songs, if I listen to twelve of their songs, it will count an equal amount of time played which is both an incorrect reflection of actual time listened but also an incorrect reflection of my preferences.

    • [Raderad användare] sa...
    • Användare
    • 2 dec 2005, 00:44

    Re:

    Quoth s3ntient:
    I don't see this being any more accurate than counting by total tracks played per artist:

    Say I have a one hour track by band A, if I start listening to it and decide after 5-10 minutes I don't like it and press skip, the plugin will have already submitted a one hour play of artist A seen as we've crossed the 4min threshhold. Now band B have 5 min long songs, if I listen to twelve of their songs, it will count an equal amount of time played which is both an incorrect reflection of actual time listened but also an incorrect reflection of my preferences.


    As I said in my last post, the plugin should submit the actual listen time. If you skip 5 minutes in, it should report you listened for 5 minutes.

    • Russ sa...
    • Alumni
    • 2 dec 2005, 00:50

    Re:

    Quoth DontLetsStart:
    It shouldn't even be that hard. Make track time a required field in the 1.2 plugin protocol (even better, make it a requirement to track total played time. If a user seeks a song it should still record the amount of time they actually listened to it!) and add a database field to capture it on every submission. For existing data and submissions on the 1.1 protocol missing it, populate with the existing data from the first time submitted. It'll be slightly inaccurate, but much better than just recording tracks and not time.


    It's a lot harder than that.

    • [Raderad användare] sa...
    • Användare
    • 2 dec 2005, 00:59

    Re: Re:

    Quoth Russ:
    Quoth DontLetsStart:
    It shouldn't even be that hard. Make track time a required field in the 1.2 plugin protocol (even better, make it a requirement to track total played time. If a user seeks a song it should still record the amount of time they actually listened to it!) and add a database field to capture it on every submission. For existing data and submissions on the 1.1 protocol missing it, populate with the existing data from the first time submitted. It'll be slightly inaccurate, but much better than just recording tracks and not time.


    It's a lot harder than that.


    Of course the devil is in the details. I meant the theory is easy. ;)

    • karit sa...
    • Användare
    • 2 dec 2005, 05:14

    Re: Re: Re:

    Quoth DontLetsStart:
    Quoth Russ:
    Quoth DontLetsStart:
    It shouldn't even be that hard. Make track time a required field in the 1.2 plugin protocol (even better, make it a requirement to track total played time. If a user seeks a song it should still record the amount of time they actually listened to it!) and add a database field to capture it on every submission. For existing data and submissions on the 1.1 protocol missing it, populate with the existing data from the first time submitted. It'll be slightly inaccurate, but much better than just recording tracks and not time.


    It's a lot harder than that.


    Plus how many media players will no longer be able to submit due to what information is alvailable via the API.

    Of course the devil is in the details. I meant the theory is easy. ;)

  • Another example

    Quoth DontLetsStart:
    Not to harp on an old idea, but looking at just the music I have on my work computer:

    Ludwig van Beethoven: 9 tracks, 5:47:25
    Dead Kennedys: 113 tracks, 5:25:05

    Well, I am harping on this old idea, just to point out another extreme example:
    Miles Davis - In a Silent Way: 2 tracks, 52:29
    Bruckner - Symphony 8 in Cm: 4 tracks, 1:22:10
    Brahms - Variations and Fugue on a Theme of Handel: 27 tracks, 26:10

    This transcends genres of music, yet one would probably notice it more with the longer tracks of classical or jazz. Seeing as our profiles are supposed to reflect accurately our own musical tastes, I admit feeling a bit put out. (These examples are from my recently played tracks.)

    Do you think your profile is misrepresented because Last.fm only pays attention to the number of tracks, rather than the length thereof? If so, join this group and commiserate.
    • muz sa...
    • Alumni
    • 18 jan 2006, 03:23
    Well, it doesn't need another person to harp on about it because;

    Russ has already said, the idea has and is under consideration, not rejection.

    This exact idea is on the "Previously Suggested Ideas" thread at the top of this forum, and it says "Read First" for a reason.

    We're all perfectly aware of varying track lengths and such, but this idea just isn't feasible at this moment in time, however, some time in the future, it may rear it's ugly head

  • Alternate solution

    I have the same problem, and I think this is an important feature for last.fm.

    We know it's difficult to implement this by tracking the song length. So why not try something else?
    Last.fm has already a perfect alternate solution. We tag music already, so why not give it length tags? Let the users specify a length classification...

    Take for instance this classification:
    - Very short: < 30 sec
    - Short: 30 sec to 2min
    - Average: 2 min to 6 min
    - Long: 6 min to 15 min
    - Very long: 15 min to 30min
    - Boring long: > 30 min

    The default value is Average. By tagging a song with lots of "Long" tags, the song falls into the classifiaction "Long".
    The beauty is, that if a song takes 7 min, the song will be probably tagged with 40% Average and 60% Long. When you calculate the average, you can get a good indication of the length.

    I think this is not so very hard to implement, and I think that users will tag songs. This because it is anoying for people to see many Short and less Long songs... Therefore people who care will tag. :)

    • AndyChrz sa...
    • Användare
    • 15 mar 2006, 22:51
    I care.

    And i won't tag. Because tagging is dumb.


    Here are the awesome record company groups that i lead here on Last.fm... Black Market Activities, Eulogy, Ferret, Ironclad, Lambgoat, Mediaskare, Metal Blade, Prosthetic, Roadrunner, Sumerian, Thorp, Trustkill, Willowtip
    • avalyn2 sa...
    • Användare
    • 16 mar 2006, 00:33

    Re: Alternate solution

    Quoth bartverkoeijen:
    I think this is not so very hard to implement, and I think that users will tag songs.

    nice idea.
    then what happens with this data - conversion of all the charts to time played rather than tracks played?

    Top Fans method keeps changing. This is the Old Top Fans formula with rationale. Discuss any Top Fans issues in the Top Fans Forum. Updated list of top obsessive fans here.
    My wife's handcrafted jewellery website www.startrishdesign.co.uk
  • Re: Re: Alternate solution

    [...]
    then what happens with this data - conversion of all the charts to time played rather than tracks played?

    Well, this will influence the relavance of songs. See it as a "multiplier".

    The numbers below are an example...
    A song classified as 'Average' counts for 1. A average Average song takes about 4 min. A average Long song takes 11 min. So, you can say: A Long song takes 3 times the time of a Average song. So the "multiplier" should be 3.

    Lets take these multipliers as an example:
    - Very short: 0.1
    - Short: 0.5
    - Normal: 1
    - Long: 3
    - Very long = 5.5
    - Boring long = 10

    These multipliers will also be influenced by the amount of length classification tags. Let's say, a song has 5 Normal tags, and 10 Long tags. This results in a multiplier of: ( (1×5) + (3×10) ) ÷ 15 = 2.3334. The average of the multipliers.

    So, if I play a song classified as Long, it will count as 3 songs in the current count system. Of cource this number should not be displayed in a times played chart, but it would be perfect to calculate a better relevance and size the bars to a better indicating length...

    So yes, "time" based charts would be better.

Anonyma användare kan inte skriva inlägg. Vänligen logga in eller skapa ett konto för att göra inlägg i forumen.