I wake up after falling asleep on the sofa sometime around 8pm and it’s now 2 o’clock in the morning. Davis is curled up on the armchair opposite and there’s just the ticking of the clock in the near darkness to keep me company. I’m now not sleepy enough to head upstairs and don’t want to disturb Is by being unable to sleep, so in hopes of tiring myself out, I tippy-tap the following…

Much as I love to check on my Last.FM page (sad, but true), there are a couple of key pieces of data that it appears to fail to capture which has the effect of failing to accurately reflect listening habits. With its focus upon artists and tracks it doesn’t reflect my recent habit of listening to compilations such as the Fonotone Records collection of bluegrass/old-time music or my own iTunes pop mixes. This is because each of these feature a large number of different artists, only a small number of which might appear way down in the lower reaches of my ‘Top Artists - Overall’ list or briefly show in the ‘Weekly Top Artists’. If Last.FM captured album information and published similar charts as it does for Artists and Tracks, it would better reflect what I’m actually listening to and make more accurate recommendations. Similarly, because it fails to register track length, Last.FM promotes my listening to the brief, piano fragments of Prokofiev’s Visions Fugitives or the blink-and-it’s-gone, thrash-core of Naked City at the cost of the drawn-out minimal techno of Fluxion or the frozen tundra-scapes of Thomas Koner. The amount of time I spend listening to the latter may well exceed the former, but the Last.FM methodology favours brevity and quantity.

While I’m on the subject, I tried out Pandora, Last.FM’s human-input counterpart (”Each friend told us their favorite artists and songs, explored the music we suggested, gave us feedback, and we in turn made new suggestions. Everybody started joking that we were now their personal DJs.”). This was a recommendation via the inestimable Dan Hill whose excellent New Music Experiences - which is just my cup of tea, though I’ve yet to finish it. However, I’m not impressed by it so far as, in response to my declared taste for Kraftwerk it suggests Throbbing Gristle, Front 242, David Carretta and Paul Van Dyk. These seem painfully US-centric in their (mis-)interpretation. Then I try creating a new station and entering Erik Satie as my first step, after indicating that he’s an artist rather than a song name I’m asked ‘Did you want the artist Erik Sanko?’. When I respond ‘No, search again’, I’m returned to a blank search box. Third time lucky, I try for Rhythm & Sound, I’m suggested Rhythm Masters… er, no thank you. Same goes for Prokofiev and Sergei Prokofiev. I have better luck with King Tubby, though I don’t like Pandora’s assessment that one of the music’s characteristics was “meandering melodic development”. Ultimately though, I don’t find most recommendations very useful and prefer to rely upon a mixture of intuition and personal research or friends’ recommendations.

[Originally published on A Personal Miscellany]

Share this post:

+

Sheesh, you turn your back for a few days well-deserved r’n'r and suddenly a giant American corporation goes and snaffles your preferred music service. When I read the news on my return, I’ll confess my heart sank a little. Last.fm’s my little secret (that I happen to share with only 20 million other subscribers). Why would they want to go and sell out to CBS? The answer’s obvious and detailed on Last.fm’s recently initiated blog. I hope they’re treated with as much of a hands off approach as Yahoo! have done with del.icio.us. I wish them a hearty best of luck and will now return to daydreaming that I’d had such a brilliant idea. Sigh.

Link: Last.fm blog post

Share this post:

We7 is a new venture, currently in beta, that offers free, legally downloadable music. The catch is that each track is prefaced with a brief (10 second) audio advertisement. There are currently only a few tracks available, but they’re from relatively well-known artists e.g. Herbie Hancock, Hall and Oates and Coolio. This may well be evidence of the sort of backing for this project (Peter Gabriel is the big name). The music is DRM-free. The We7 website states that relevant ads will be attached to the track on download so that they can be targeted according to user demographics.

When I registered with the website, the technology worked smoothly and I was able to download and play a Herbie Hancock track without a problem. The advertisement consisted of a female voice stating “I’ll pay for a download when I die, until then… We7″.  I’d rather not have had to listen to it, but it was brief and pretty painless. The business model is certainly innovative - it will be interesting to see whether it succeeds when it launches properly this summer.

Link: We7
Via: Soundtracking

Share this post:

That’s a question that’ll quicken the pulse of many a music geek. Am I narrow-minded or open-minded? Am I just right? What is right? Okay, I’ll rein back on the worrying, but I have wondered whether the plethora of music recommendation services will result in more or less homogeneity. Anthony Lieken’s script isn’t going to provide the answer, but - if you have a Last.fm account - it does promise to tell you how eclectic your listening habits are relative other Last.fm subscribers:

The following script takes the 20 top artists in your musical profile from Last.fm, and finds the collection of top 5 similar artists for this top 20. The resulting is a list of artists similar to your preferred artists. As the list is larger (maximum = 100), your musical preference is more diverse.

If this tweaks your curiosity, the author has a number of other online scripts that might be of interest.

Just in case you were wondering (as I was), how eclectic Music Interfaces is - I got a 90 out of 100 (I ascribe that to my love of Barbra Streisand and Bad Brains - er, I’m joking about one of those…)

Links: Anthony Lieken.net, Last.fm,

Share this post:

Click for full size image version

Visualising information is a fascinating area and, as much of a fan as I am of Last.fm, the user charts tend to be a pretty uninspiring sight. Long-term users’ charts like my own also tend to be fairly static. If Last.fm were to employ the services of Lee Byron, things might be rather different:

Algorithmically generated posters based on statistical information provided by Last.fm software. Every song listened to by a particular user over an eighteen month period of time is recorded and used to create the visualization. Each colored band represents a musical artist, progressing left to right through the eighteen month span growing wider when listening was more frequent, and skinnier when it was not. The hue of the artist represents the time of the first listen for the particular user: cooler colors represent artists who have been listened to for a long period of time while warmer colors represent artists who are more recent in the user’s listening habits.

How nice to have your Last.fm personal homepage with this information presented in Flash allowing you to interact with all that data and/or project it as a gradually morphing illustration of the music you’re listening to… With at least 15 million users, Last.fm would need to do some serious hardware upgrades, but it would be music geek heaven. Following links from Lee Byron’s initial page leads to a rather lovely interactive graph of Artist Popularity Over Time (warning: Applet, which seems to only intermittently play nicely with Firefox on OSX):

Click for full size image version

Links: Lee Byron/megamu, Last.fm, Information Aesthetics
Via: Mediaor

Share this post:

Is there no stopping Last.fm? Following on from recent news of partnerships with some of the top four major record labels comes news of a global chart show. All power to their elbow, though I’m not sure Channel 4 is currently a big player in radio:

The Worldwide Chart will showcase the top tracks from Last.fm’s European, Asian and US charts. The station currently has 20m active users from over 232 countries and is available in 11 languages. Channel 4 Radio DJs will also begin ‘scrobbling’ their playlists, allowing Last.fm users to track and share their selected music.

Link: New Media Exchange article, via Digg

Share this post:

Or to put it another way, how underground are you? Fun for Last.fm subscribers. I eagerly entered my own Last.fm ID and learnt that I’m 10.01% mainstream. Should I be overjoyed, disappointed or indifferent? I was apparently “compared to an average of 530447.1 listeners in the Last.fm top 10″ - I have no idea what that means by the way. I’d certainly encourage you to click on the links to the most undergrounders users - some of the artist names are brilliant (Chocolate Billy, The Chicken Masters or Cosmic Wurst anybody?)

Link: Mainstream-O-Meter
Via: Digg

Share this post:

Not so long ago half your record collection would sit in the attic, gathering dust. In 2006 it’s more likely to be languishing in the dark recesses of your iPod or laptop, unplayed, unloved. We always think we’ll get round to making great playlists and rediscover hidden gems, but let’s face it, who’s got the time?

So declares The Filter’s homepage. Until last week this free software was only available for PC, but a new Apple-compatible version has just been released. The Filter’s profile is probably significantly higher than its market share due to the frequency with which it’s associated with Peter Gabriel’s name (certainly did the trick for me). It turns out that he’s an investor in the project rather than having much to do with the product’s creative approach.

The Filter acts in partnership with iTunes on Windows XP and OS X (a Windows Media Player version is in development) to:

• Create playlists from a user’s existing music library using user-specified seed tracks.
• Making recommendations, again based upon tracks suggested by the user.

The playlist function is very like the primary purpose of Music IP Mixer, a product I looked at a couple of months ago. Like that program, The Filter is standalone rather than web-based (like Last.fm, Mog, Pandora) or an iTunes plug-in (as per iLike). The main window is an attractive shelf design on which are displayed the upcoming cover graphics for the songs in its active playlist:

An additional window, which can be closed, also opens up with good quality information about the playing track supplied by the AllMusicGuide. By default both windows stay on top of all other open programme windows, which can be rather awkward. The window can be minimised and reopened using a keyboard shortcut, though this still doesn’t make it quite as user-friendly as iLike’s iTunes drawer. It would be good to see the window accessible via its OS X statusbar icon, but such an option isn’t currently available.

To create a playlist, the user chooses three different tracks in iTunes and clicks the ‘F’ button in the player window. My first three seed tracks (Brian Eno’s Energy Fools The Magician, The Durutti Column’s The Beggar and Manu Dibango’s Massa Lemba), chosen on impulse, resulted in a playlist of 15 tracks with five suggested artists and the rest made up of additional tracks by the chosen artists. The suggestions (Peter Tosh, Fela Kuti, John Martyn, Laurie Anderson and U Roy) were reasonably on target. Given the aforementioned proximity of The Filter’s functionality to Music IP Mixer’s, it seemed a good idea to try it out with the same seed. Choosing Erik Satie’s Fantaisie-Valse, Harmonia’s Kekse and a part of Steve Reich’s Three Tales multimedia opera resulted in a different, but still serviceable playlist that included Sufjan Stevens, Vladislav Delay and Bach amongst others. Skipping songs quickly or letting them play through enables The Filter to build a better idea of listening habits. The programme can also build playlists to sync with iPods - these are based on either a chosen genre or, as before, by selecting seed tracks.

The recommendations option produces a window of recommended songs available at the iTunes Music Store:

Some of these recommendations are good, some are not - how Nirvana’s Smells Like Teen Spirit relates to Brian Eno’s Another Green World is anybody’s guess… A more mainstream track (Madonna’s Confessions of a Dance Floor) results in a fairly diverse mainstream set of recommendations (Sean Paul, Britney, Cyndi Lauper and Beyonce) which it’s hard to argue with. Clicking any track opens up the iTunes Music Store to listen to the standard 30 second clip and purchase if so desired.

Recommendation: My copy of The Filter tended to hang fairly frequently and was slow at delivering iTMS recommendations, but the program is still in beta so perhaps that can be excused. It’s an interesting product, particularly for its integration with the iTunes Music Store which makes for potentially intuitive discoveries - the inevitable downside is that you’re locked into Apple’s DRM. Its ability to offer up forgotten gems from an overstocked iTunes library is also attractive. In my experience, its recommendations have been fairly mixed, but The Filter is definitely worth trying.

Link: The Filter
Link: Screencast guide to The Filter

Share this post:

In my recent look at Mog, I found the social music recommendation site (gosh that’s a mouthful) good on the social and less impressive on the recommendation side. They announced the other day an improved service in the latter category, so here’s a quick look. Acting entirely in character, Mog’s new recommendations don’t just appear in a long list, but are tied to specific users:

See that round ‘magic button’ on the left sidebar? Clicking it does this, accordng to Mog:

If you are running our MOG-O-MATIC software, The Magic Button will show recommendations from the moggers who most closely match your musical taste. We can determine which moggers share your taste by comparing your collection and listening habits with those of all other moggers who are running MOG-O-MATIC.

Looking at what I thought were the Mogs’ recommendations led me to think their accuracy was way off - I’m just never likely to be a fan of Keane, Muse or The Killers. However, it turns out the default listing on display is ‘Recommendations: Last Songs Played from Featured MOGs’. So is the magic button really magic? Oddly, clicking it resulted in no change to the page: same list of featured Mogs, same songs which only changed if I chose a different breakdown e.g. top songs this month or top albums this week. This way of making recommendations just doesn’t seem very intuitive; I don’t know these people (’featured Mogs’) and it’s not clear how their tastes relate to mine. It may be that Mog requires users to be more sociable than I have been, more proactive in making friends and interacting in order to increase the accuracy of recommendations. If that is the case, this seems more of an onus than gradually building up charts by playing music that Last.fm requires.

My Mog homepage still contains a box prominently displayed at the top of the page ‘Artists you should know about’ and the instruction ‘Click edit to add some’. I guess I can safely conclude that I’m expected to add artists I think visitors should listen to - but I still think this isn’t very intuitive to new users. Overall, these developments don’t seem particularly useful, at least on initial impressions.

Share this post:

iLike is one of a growing number of services that offer to make recommendations about music you may have not have heard based upon the music you’re listening to. The vast majority of these services (Pandora being the main exception) function by comparing listener’s tastes with each other, and making suggestions based upon statistical analysis. At the same time, being all Web 2.0, these companies are keen to offer some form of social function based upon shared musical tastes. As iLike claims:

Check out what your friends are listening to, browse the libraries of people with similar tastes, and get Free MP3 downloads by new artists matched to your own music tastes… Share music libraries with your friends, browse and sample their most played songs, and compare your compatibility scores… Auto-organize your iTunes library, share your music tastes with friends, and get music recommendations directly in iTunes.

Although this is very similar to Last.fm and Mog’s offerings, iLike does things a little differently in terms of its interface. This is because, rather than the standard web-based interface, installation of the free download results in a drawer appearing to the right of the main iTunes window, as illustrated below:

The drawer has its pros and cons. It’s very convenient to be able to see music recommendations for the artist currently playing. If iLike is able to propose similar music for a particular artist, it will list up to five tracks in the upper half of the drawer together with ‘free music by new artists’ in the lower half. Each track has a little play button beside it which automatically pauses iTunes and plays a 30 second sample. Clicking the arrow to the right of the listing takes the user to the relevant track on the iTunes Music Store, if it’s available. Two other tabs provide quick access to any Friends you may have made on iLike’s website, as well as providing the ability to customise your music profile and accessing the user’s iLike homepage for messages, recommendations, etc.

A user’s iLike homepage provides a fairly standard array of lists and options including:

• Recently and most played songs
• Friends
• People with similar tastes
• Playlists (in early preview form)

All of this is cleanly and intuitively executed. The disadvantage of the iTunes drawer is that it uses up valuable screen real-estate, whether or not this is an issue will depend upon the resolution of your monitor. On my Macbook Pro (1440 x 880 pixels), it’s just about okay. As stated earlier, recommendations are dependent upon analysis and comparison of other users’ listening habits. This is where iLike is less useful and often less accurate because they clearly have a much smaller database of user’s tastes. Although iLike was reasonably good on a fairly wide range of music including, pop, rock, acid jazz, jazz and funk amongst others, it did fail to come up with any suggestions for less mainstream music. The screenshots below are variously for electronica, world music and techno artists:

As iLike’s user base grows, this will gradually improve and the usefulness of the interface is likely to persuade some potential users to try the service out.The sidebar is stated to work only with iTunes on Mac OSX and Windows XP, Vista is in the process of being tested and a Windows Media Player version is in preparation (currently it tracks usage, but lacks a user interface).

Recommendation: iLike is well worth trying although the most accurate and in-depth recommendations remain the preserve of Last.fm. The social side of the equation is probably better served by Mog. However, the instantaneous nature of iLike’s recommendations - where they are available - is a definite advantage.

Link: iLike
Link: MusicInterface’s iLike homepage
See also: Overview of music recommendation services

Share this post: