What do YOU want in ListenBrainz?

ListenBrainz could provide charts based either on the number of listens (like it does now), or charts normalized by listening time. This could be selected by an option in user preferences.

Last.fm is unable to do this, because they don’t have lengths for a lot of the tracks, but LB is in different situation given the integration with MusicBrainz database.

User’s tracks with unknown lengths could be collected in a separate page, and presented in a similar way as tracks not linked to MB recordings, so the user could add missing data.

6 Likes

there’s a ticket for it now, but adding playlist covers to ListenBrainz is pretty high on my wishlist~ I quite enjoy creating such covers myself, and perhaps that could be built into the website too? (see LB-1215 as well)

edit: these could in theory allow users to synchronize playlist covers between multiple libraries and services? or at least be a singular place to keep them…

1 Like

Just glancing at this thread, so forgive me if this idea has been mentioned already, but to make ListenBrainz more engaging to actually browse as a site/dataset, I think it should try to be ‘smarter’ about pulling URLs for its Play button. From what I’ve seen, all the Play button seems to do on LBz is to pop open a youtube embed of the first search result… Anyone remember the days of the “I’m feeling lucky” search? :laughing:

For example, there are a lot of MusicBrainz Recordings that have “stream for free” links that go to youtube videos (and for that matter, a lot that go to other platforms or even a direct download-for-free link). For the most common of these, I think it’d be relatively easy to tell when these are available for an MBz Recording entity, and present those as the Play stream.

I can understand if LBz is hesitant to do this because it could make the site look sketchy, rights-wise, since making free streaming access to certain music ‘too easy’ will earn the legal ire of big labels and such, but if the LBz team isn’t too worried about that (they added the Play button in the first place :thinking: ), then this is an augmentation to it I’d love to see.

5 Likes

for what it’s worth, the YouTube links linked to MusicBrainz recordings should (whenever possible) be the official video (or the video for YouTube Music)

1 Like

Feel free to vote this up!

1 Like

This is a big post, so please forgive me. I’ve used last.fm since 2016 so I’ve had years to think about what I would like to see in sites like this(and what I liked most on there). And given how LB is linked to MB it right away solves the biggest problem last.fm has refused to deal with since it was created(grouping artists by name, and often redirecting things wrong and straight up losing months of stats for them because of the way they do it). So this is my thoughts after years of last.fm, and a month of having LB open all day every day. Forgive me if I mention anything others have said, or if I missed a feature that already exists. (I don’t have stuff in year in music, or recommendations, or fresh releases, etc so I’m not sure what they all will look like)

Tracks, albums, and artists should have pages on Listenbrainz

that you go to instead of musicbrainz, similar to how last.fm does it. I would like to see similar info on those pages that last.fm shows. Play counts, view your historical listens, a graph of listening trends from all users(I love these especially as it’s neat to see the spikes for artists when they release new albums, or when certain songs cause spikes at times of year like christmas or halloween for music appropriate for those times). Also it would be nice if you could see the total number of listeners, and total listens for each item like on last.fm.

Musicbrainz allows linking socials for artists, so on a listenbrainz artist page it could have a social feed showing recent posts by that artist. Is there maybe a 3rd party feed service with an API that takes a list of social links and provides a feed that could be integrated without adding extra server load?

It would also be nice to have a list of similar artists on a listenbrainz artist page. I use this a LOT on last.fm as they do a good job of showing similar artists, not just popular artists from the same genre.

Social features. Comments, DMs, sharing

Last.fm shouts on song pages I feel are 95% useless as harsh as that may come off. And they are often downright awful and should be removed by moderators. Instead I would love to see Critiquebrainz reviews embedded without leaving LB. But shouts for users and artists are better(than they are on songs, not vs critiquebrainz) in my experience. Especially for users as it adds to the social aspect and makes the site more engaging I feel. And for users you could have it so they can delete any comments on their page. And for artists either downvote threshold or report for bad ones?

It would be nice to have a DM feature to message other users as well. But have it so people can pick between disabling it completely, or limiting it to people they follow, or that follow them to lower spam and abuse(I’ve gotten spam in my last.fm DMs).

Also for social features, a small bio + profile picture that would be displayed similar to last.fm would be nice, I find it helps weed out real people vs bots and gives you an idea about a user. Social features like that give people more reason to spend more time on the site and encourage others to use it, I think, rather than just once a month or two looking at their stats.

On the topic of social stuff. Maybe more ways to share some of the pretty graphs you get on social media beyond a link. Something you can embed in posts. Would be a good way to get more users if people start sharing their stats for others to see.

The site already shows you users you are similar to. It would be nice if you could see a score for how similar you are to any user when you view their page.

Loved Tracks

One difference with Listenbrainz and last.fm is how Loved tracks are shown. I prefer how last.fm does it when viewing other users profiles. On Listenbrainz when viewing another users profile it will show hearts for songs I have Loved. On last.fm it shows tracks that user has loved. I feel the way last.fm does it is better for browsing other peoples listens.

Recommendations. Diamonds in the rough? Quality > Quantity.

I mostly listen to kpop these days. And with KPop especially, but other genres as well, how popular something is all comes down to marketing. Even social media trends are all marketing. So to me seeing some group has 2 billion views on youtube means little to nothing on how good they or the song is. And something I noticed browsing last.fm often was a lot of the trendy groups had much worse ratios for total listeners to total listens. Lots of people listen a few times and that’s it. Compared to smaller groups they often had much higher listens to listeners. And some songs in particular had much higher listens to listeners ratio. So what I would like to see is highlighting music in this way. Finding songs and groups similar to what I like, that people who listen to them REALLY like them. Less about total popularity.

Similarly. Last.fm shows a graph on every artist, album, and song page showing listening trends. And again a lot of really trendy groups and songs you will see the typical increase when a new release comes out, but then a rather sharp drop after. But then some groups will have much slower decrease, or even a constant increase in listeners over time. So it would be nice to highlight releases like this, that have a higher… staying power? Again not showing what is trendy, but stuff that people continue to love over time. An example here would be looking at “miss A” a group active from 2010 to 2017 shows a gradual increase in listeners over time despite no new music, while “IVE” a new group shows a spike at their last release, and then a constant downward trend since.

Again speaking to my current taste in KPop. There is a lot of amazing and talented groups out there, who make music of the same style and quality of well known groups, but have much smaller companies behind them. So they aren’t marketed as well and people often never even hear about them. I would love to see more recommendation features that try to not just recommend the well known groups, but everyone regardless of how well known they are. Like, no one interested in KPop needs to be recommended BTS, even people who don’t like KPop likely know them. Find me the artists that have gone under the mainstream radar.

Obsessions

I listen to a lot of different groups. I have a playlist with a few hundred groups on it as my main list that I play on shuffle. But some times I will come across a new song, or even one I’ve known for awhile and can’t get enough of it. If one was to analyze my listening history they’d probably find quite often there will be sudden times where a single song can be played anywhere from 5 to 150 times in a row(Literally, I’m a bit crazy). I would love a feature that could find these songs from all users and rank them again in a way fair to all artists no matter how big or small. Like number of obsessive streaks relative to total listeners or something. And then show me ones similar to music I like.

But also have a list of obsessions like this on peoples profiles so you can see what your own, and other peoples obsessions are. Not just what I say I love. Expose me for listening to Babyface by WJSN for 16 hours straight.

I would love to see more general(not user specific) playlists like spotify has, or like Plexamps auto playlists.

Using data from musicbrainz and listening data from listenbrainz generate weekly lists and let users save them permanently if they like them. Could be the top trending songs from a country, or genre. You could either use detected locations or allow users to input their City/State or Province/Country and show music people near them enjoy. Could be a time travel playlist for a genre with popular songs spanning a range of time. For time travel maybe variable length and start/end, like Could be longer and have top song from each month over a period, or shorter with just the top song of the year(in that genre).

Related to how I mentioned spikes in last.fm graphs at certain times of year. You could analyze your data to generate playlists for holidays. These songs are popular on Christmas, or halloween, or valentines day, etc. Find songs that spike specifically at those times of year. (For example look at the graph on last.fm for my favourite christmas song https://www.last.fm/music/PURPLE+KISS/_/My+My).

More ways to show top artists/albums/tracks?

It would be cool if the site tracked your top artists, albums, and tracks for each week, then gave you a list showing ones with the most time as #1 in the lists. Would be different from total plays as I could listen to one song 300 times a day for a week and have it at the top in my total listens, then never listen again, or could have 50 listens a week and be #1 and have that happen for 30 weeks straight.

It would be nice to see things like your Top New Artist for artists who debuted that year, and Top New Discovery for artists who aren’t as new, maybe Biggest Gains? for artists you heard before the current year but moved up the most in your top artist list?

Issues? Bugs?

Groups some times show up on my toplist multiple times, even though they have the same artist ID linked. “9Muses” and “Nine Muses” for example are my #20 and #21(Using them as a good example as they just happen to be next to each other, happens a lot though). It would be nice if these would be get automatically merged. It could be especially bad with KPop as groups will have releases credited in many different ways. Having both English and Korean for the same release depending on the source(spotify vs bugs! for example) are very common, plus Japanese for Japanese albums, Chinese for Chinese albums, etc.

I haven’t looked to see this, but seeing those like that, I imagine if groups change their names it will also have multiple entries for those even though it’s the same group.

Likewise, it would be nice to see sub-groups merged into the main group. They have unique pages on musicbrainz, but I do see they have the sub-group relationship, so it would be nice if it would find an artist is a sub group of another it would merge into the main group on listenbrainz stats pages. Maybe it could be expandable and show each main/sub/member under there. Like Girls’ Generation over all 3000 listens, expanded to Girls’ Generation main group 2000, Girls’ Generation-TTS 500, Girls’ Generation-Oh!GG 500. Or the bar sectioned to show the amount each sub group is responsible for and it shows the name on mouse over. Cause in the end its still the same group, just a subset performing that song. If a rock band does an acoustic song and some of the members sit out you don’t call it a different group do you? KPop having their own subgroup names is all just marketing to create buzz.

Also with featured artists the top list will show everything as it’s own entry rather than grouped. So for say an artist like RIELL no matter how much I listen to her she’ll never show up anywhere near the top of my list, as the majority of her songs have different collaborators. For a person like RIELL I get that it might be tricky on how to display it when its shared credit. But I’d rather things not add up and have double credit (like both RIELL and Besomorph getting full credit in the top list for songs they collab on) than crediting each way the artist credit was written separately and breaking up the listen counts for artists, as if you look at RIELL she has been credited in 25 different ways.
These various things for me at least kind of take away from the top artist list as I feel like they don’t get the credit they deserve?

Also related, but for the world map, if there is more than one artist for a track, it does counts listens for all artists towards the total for the country it seems. So for a song like Not Friends by LOONA, rather than the group getting credit, it credits 4 different members of the group, so each listen of that song gives 4 to the total for South Korea. (I noticed this when I only had 1 song for a country and it had a featured artist and the total was the combined total. 20 listens of one song gave 20 to artist 1, and 20 to artist 2, and 40 total for the country. Not a huge deal, just can lead to inflated listens on the map.

Data missing tools

I would like to see more “Data missing” tools for music I listen to. Not just for stuff that is missing completely, but it would be nice to see a list of artists I listen to who don’t have an area set for example(I’m obsessed with my listening map). It would also be nice to see on the world map an indicator how many listens/artists aren’t being mapped. I have been trying to add missing artists, releases, and links for stuff like streaming pages and socials, but it’s time consuming just checking hundreds of artists musicbrainz pages to see if they are missing stuff like that. I feel like exposing more info in ways like this would encourage more users to add missing information.

Musicbrainz collections + Listenbrainz

As an example, there is a collection on musicbrainz of hundreds of different KPop groups. I like to try and find music I like regardless of group popularity. It would be amazing if you could take collections like that, and have a randomized playlist of top songs from each group in the list. “Play collection on Listenbrainz” from MB? Looking at users on MB I have seen some people have sizable collections they have organized so it would be neat to be able to play them like this.

Linking and auto matching

The link UI needs a rework I think to make it more user friendly. Personally I am obsessed with my stats and go through the effort of looking up each recording on MB and copying the recording ID. But looking at my similar users, many either don’t know about it, or don’t want to spend the time as they have a lot of unlinked songs(That I know exist as I have linked them on my own page). When adding releases to MB you get a nice box(that could be autofilled with the information from that listen) that pops up suggested artists, then again for release and recording. Seems like the code for a more user friendly experience already exists and could be brought over to LB.
Automatic matching also seems to struggle with releases that aren’t only in latin script. A lot of my KPop songs for example fail even if the artist, album, and song are in latin script, when there are multiple recording names in MB that also have hangul names, even if there are some that are perfect matches. Almost seems like it doesn’t compare to ALL versions of a recording and just the default one?

Also, it would be nice if there was an easy way to report incorrect matches. While I can fix my own, it doesn’t seem to help with anyone elses. The Queen of KPop as an example. “BoA”'s track “Forgive me” off her EP “Forgive me” matched as Zoya’s Forgive & Forget(I think. I know it’s that artist as it’s on my similar users map linked still, I fixed mine long ago though)

Remember last.fm username

I was looking through my missing data page and started coming across really… different… artists than I listen to. Then one day I clicked to the last page of my listens and seen even more strange stuff. Then realized the dates were 7 years before I started using last.fm. So I assume at one point I accidentally entered the wrong name into the importer and it grabbed someone elses, oops. So I had to delete my listens and reimport them to clean it up. So having it remember usernames would be great to prevent accidents like mine.

Help? FAQ?

Another addition I think would help would be more help pages, or help tooltips, or something. For example, on the reports page it would be nice to have a help or FAQ that explained how often all the data is updated. Likewise for any pages that aren’t real time. I atleast can’t seem to find much, though I could be just missing something right in front of me… again.

Also if there is ways to monetize this site in a user friendly way to help cover costs and provide more money for more server power I would support this. Even if it was deals with services using affliate links or whatever. Buy this album on Amazon, or Apple, etc. Are non-profits able to do stuff like that?

Edit to add: Dark Theme

Thank your for all your hard work on this site and musicbrainz! It is especially refreshing to see you engaging with users for ideas in contrast to last.fm that has ignored major issues for… ever. I wish I was smart enough to help as a developer. Hopefully my obsessive listening data and (my other accounts) MB contributions help!

10 Likes

I second most of this, also answering a few points

I can tell there’s work being done on this, as there are very barebones release pages (like this) if you click the “Play on ListenBrainz” button right below the cover art on a MusicBrainz release. I don’t think there’s any way to access them directly from ListenBrainz at this time, but this is a feature I’m very much looking forward to~

they did something like this for this year’s Year in Music (see the beginning of this thread or my Year in Music), and based on some of the mockups @aerozol has been putting together, that’s in the minds of the design people. when and whether the developers get that in might be another story, but we can hope :wink:

this is a great idea, actually. and (in my mind, note I’m not a developer) might be pretty easy to implement. the ratio of listeners to listens I think could be a great indicator of what people like, no matter how popular a track or artist is.

I am kinda obsessed (lol GET IT) with this whole idea, and it also reminded me of one of Spotify’s billboard campaigns back in 2016. such an excellent idea~

I know this is something being worked on in general (https://tickets.metabrainz.org/browse/LB-1192), and two of the cases you’ve brought up here are probably already covered by LB-1191 and LB-631

this feature was pretty recently added, and I don’t think it has been publicized much outside of the forums, the MetaBrainz blog, and maybe IRC (not sure, as I’m not on the IRC).

I believe there’s a ticket for this too (LB-1036), and you can match other user’s listens as well (more details in the comments of LB-1198)

3 Likes

Thanks for the feedback @MusicListenzGood! Please check out ListenBrainz Redesign, particularly the figma design project linked within, for where we’re headed with the project. I’ve already made comments in the figma project with a lot of your feedback, but please feel free to use the comment function to reply or add new comments.

I’ll address some things here as well, if I haven’t quoted something it doesn’t mean it has gone unread though :slight_smile:
Everything @UltimateRiff added was spot on as well, of course.

This is a big chunk of feedback at once so please don’t hesitate to bring things up again later, bump people, or create tickets or threads for specific parts of this that you would like to get more eyes on.

—---------------------------------------------------

Definitely in the pipeline, love your suggestions though. Please keep an eye on the design thread as I haven’t mocked these up yet at all, and feedback will be welcome.

The decision was made at the last summit that the CritiqueBrainz frontend will take the lonely road (be shut down*) after all of its features have been migrated to MB/LB/BB, so the ‘reviews in CB’ is definitely going to happen. *CB lovers, don’t worry, it’s not going anywhere soon, will be lots of time for discussion when the time comes

I (and a lot of the team) used to be a big user of last.fm, so I was pushing for shoutboxes for a while, because they were a big part of my experience there. However MetaBrainz staff with more ‘web 2.0’ experience said definitely not, and when we recently had meetings with a former last.fm developer/team lead they also said ‘definitely not’. So I have come to believe that we can give people good, satisfying, ways to interact around music without resorting to a global shoutbox on entities. Not sure what form that’s going to take (though we already have review comments, and ‘recommend track to’, and ‘pin recording’, and so on, which allow free text), but I think it’s fair to say: don’t expect shoutboxes in LB.

There is a bigger dev discussion floating around around how to do a notification system, which DM’s may (but, same as shoutboxes, very possible not) be a part of. Probably this notification system should be a global MeB one. For that we need to finish the universal MeB universal system. And so on… basically… don’t hold your breath. But also don’t despair!

As mentioned, we tested a lot of this in the last ‘year in music’, and I’m keen to expand on it.
Ideas welcome! There are lots of LB ‘grid’ mockups here, but that is just one way to share things. Recently I’ve also realised that we should give artists a way to share their stats/fun things as well (they drive a lot of new traffic/their listeners to Spotify et al). and so on.

Last.fm already did a lot of work around this. An example that came up in our meetings (from my notes) was “A user listening to a popular artist once isn’t really valuable data - solved by taking the top listeners of an artist to compute weights initially (so Thom Yorke appears in Radioheads top related artists… even if most people who listen to Radiohead just listen to other ‘popular’ stuff)”

I’m sure there are heaps of other things to take into account. @lucifer and @rob have been working with this, mainly in regards to creating playlists, but I don’t see why it wouldn’t translate into any ‘similar/recommended artist’ features we eventually do.

This is the kind of cool stuff I would really like to put into our ‘feed’ page, which currently is just literally a feed of some listens from your follows. You can see my mockup in the figma. The mockup is fairly conservative, but it starts to hint at tying in what other people/the global community is up to (X is trending), and what artists are up to (X released an album). There’s so much potential for this to be the hub for music updates!

btw, re. social feeds,it’s so hard to display things from other services (e.g. the big companies) without jumping through a lot of unclear (and changing without warning) hoops. Even if the technical implementation is possible, easy even, anything that requires calls on other services is ultimately up to how open their API is.

@rob, @lucifer, what about something like this? I could mockup something based on our other ‘explore’ pages, with some sliders and options, similar to the grid ones. Since you’re neck-deep in playlists at the moment maybe this interests one of you.

Check the rest of MusicListenzGood’s post above for some more interesting thoughts under ‘I would love to see more general(not user specific) playlists…’

I think once we expand the data we show in LB (two things I’m thinking of in particular, the artist/album/track pages, but also a drop down ‘info’ panel on user pages, for listens, which we’re planning), it will be very quickly obvious when data is missing. Then we can put little ‘edit’ links in the gaps. Same for tools that use artist country, genre, and so on. Because we’re not surfacing that data that much at this early stage it hasn’t come into play yet.

I fully expect every page on MB to be playable in LB eventually, I don’t see why we wouldn’t. This is also a dream of mine (in addition to being able to see what you’ve listened to/a LB listen count for every entity on a MB page…)

You’re not alone, please add a vote/comment to https://tickets.metabrainz.org/browse/LB-1040

This is something you would be very very welcome to contribute to, if you have the time and the inclination to sketch out a bit of a plan, and then start filling in details :slight_smile:

3 Likes

Yes, that is on my radar already and not that far off. We’ve been testing out the Daily Jams playlists that get generated on a daily basis to get a feel for how our collaborative filtering is working. We’ve been slowly tuning it and it is producing some pretty good results. To get playlists made for you, follow the “troi-bot” user on listenbrainz:

https://listenbrainz.org/user/troi-bot/

Then, if you have recent listens and have set your timezone correctly, shortly after midnight in your selected timezone we will generate the playlist and place it into your collaborations tab on LB. (mine are here rob's Playlists - ListenBrainz )

We’ll build a real UI for this before too long, but consider this to be an “unreleased feature” for now. The feedback we’ve gotten is that “this is good, but I would prefer a weekly playlist”. Yup, agreed – am only doing this daily, so I can learn on fast forward! So, very soon we will have “weekly jams” (feel good, easy to listen to) and “weekly new jams” (new tracks that you might like). All of these will be delivered as playlists, so the user can clone them during the week to keep the list. Otherwise the list is replaced the next week (or tomorrow with daily jams).

Also on our radar is Artist Radio, where a user can give a few seed artists and the system will dish up tracks for you, similar to the wildly popular last.fm feature. In order to make this feature happen, we are working on Similar Artists, Similar Recordings, Popular Recordings datasets (which is quite a lot of work as well). Once we have these datasets matured sufficiently (we’re close!) we’ll expose these as features to our users.

@aerozol has done a great job of laying out everything we’re up to – thanks for that. I wanted to add a bit more from a high level perspective: We’ve build a LOT of data tools over the past 5 years and we now understand what we’ve built and are ready to be more UI focused for a while to bring all these new features to light in a redesigned UI. Some of the early steps of this are already in PRs that will do some really necessary cleanup, before we dive deeper and generally improve the whole look and feel of the entire site.

Thanks for all the feedback, this thread has been been very helpful for us!

6 Likes

Is there something more I have to do? For several days now I am following troi-bot but I don’t see any new playlist in collaborations. Recent listenings should be sufficient. The time zone was never a problem, my listenings are showing the correct time.

it looks like it hasn’t been making playlists the past few days, I got my last one on the 9th

1 Like

ooh, I’ve just thought of another feature that would be great, to have some sort of album compatibility score or something along those lines. might be better with a use case example

say I’m at a secondhand music store. I pick up a CD I’ve never seen before, but the album art looks pretty cool. I look it up on ListenBrainz, (maybe by barcode or cat. number?), and see that based on my listening history, there’s a 68% chance I’d like this album. I pick up another album, and there’s a 24% chance I’d like it, etc. perhaps it could even look at your MusicBrainz collections to see if you already own it?

this would be a really nice feature for those of us who still do physical media, like myself

4 Likes

Oops, sorry we had a bug and then a deployment snafu for our daily-jams (which is why this isn’t an official feature yet, heh). We’ve now fixed the issue and I’ll see if we can run daily jams for everyone now.

Stay tuned.

2 Likes

Oh, quite interesting. I’ll have to think about how I could build such a feature.

2 Likes

OK, we fixed the backend now. In order to get daily jams you have two options:

  1. Wait until midnight in your chosen timezone – shortly after midnight a playlist should appear in the collaborations tab.
  2. Or, if you’re really set on getting it done sooner: set your timezone to a timezone where it is about to be midnight. Then wait for the playlist and set the timezone back. :slight_smile:

(Currently the timezones are only used for determining when to run daily-jams, but that will change soon).

2 Likes

I think I brought this up very briefly when we were working on the Fresh Releases mockup. If that’s got ‘compatibility scores’ for individual users why not display them everywhere? My only thought was that the way we currently calculate it (assuming that if the score is 0 it doesn’t show on Fresh Releases) it might end up with 0 for 99% of the DB.

I assume for Fresh Releases we just check “how much have you listened to this artist already”? But could combo or replace that with how we generate playlists.

Related to my guessing at that, I would love to know why something’s getting a score. Because I listened to x + x a little or a lot, or the same genre, and so on. The Spotify ‘black box’ method isn’t super interesting to me.

Thank You so much for all this work! It’s awesome how ListenBrainz was developed and is developing at such a rate.
CONGRATULATIONS!

4 Likes

The problem is that the fresh releases score is how a given user relates to this release. It is not a score of how a release relates between two people. That’s a totally different data-set that we would need to build.

I think it is a great data-set to have, but it isn’t trivial to generate. I think the best thing we can do for the time being is to make a ticket for this suggestion and let us get some of the many open tasks worked down before diving into this.