Duke Listens!: Visit my main blog at MusicMachinery.com

Apple has just rolled out the 'Complete my album' feature for iTunes.  With this feature, Apple will give you a $.99 credit toward the purchase of an album for every song off that album that you've purchased individually.  This lets you easily try out a few songs from an album  and if you like them you can pick up the rest of the album without paying any penalty.  It's a super feature, that will hopefully encourage listeners to move beyond the most popular tracks for an artist to the tracks that may take a while to grow on you.  I think Apple should go one step further with this and give you a discount for each additional song you buy on an album: $0.99 for the first track, $0.88 for the second track, $0.77 for the third and so on until you reach $4.95 which should be the cost of the album.

This morning, I was fixing a bug using Netbeans.   Typical cycle: think, type, build, test.  However, the changes I made didn't seem to fix the bug.  I was nearly 100% sure that I'd identified the problem, but running the test was still showing the failure.  So another cycle of think, type, build and test.  Still, no dice.  Time to add some printlns - think, type, build, test. Woah! none of my debugging prints are showing up - what's going on? Dang!  The project I was working on was not my 'main' project in Netbeans. So when I hit build, it built the wrong project, so my changes were never being compiled. 

Netbeans should be smart enough to know that when you have multiple projects open and you are editing some source and you hit build you are likely to want to build the project related to the source that you just edited, and not the project that has no changes.  I'm probably bit by this at least once a week.  You'd think I'd learn my lesson, but I don't. Sigh.

There are many web sites on the net that know what I like.  Last.fm knows what music I listen to. LibraryThing knows what books I read.  The findory knows what  type of news I like.  Google knows what kind of searches I make.  All of this attention data really should belong to me since this is data about me that is generated by me.  If I go to the iTunes store I should be able to point them to my last.fm profile so that iTunes can generate better music recommendations.  If I go to Amazon, I should be able to carry along my taste in books based on my LibraryThing collection.  I should be able to, but right now, I can't.  iTunes doesn't know about the 3 years of music taste data that I've accumulated at last.fm, and there's no way for me to tell iTunes about it.  What we really need is a way to make our 'attention data' portable - so I can take the data about what I like and take it with me.  So when I go to a new music store I don't have to start from scratch describing what I like and don't like.  Instead, I can start getting good music recommendations immediately based on my years of accumulated music taste data.  Of course there are lots of issues here - privacy, security and portability don't always mix well.

The Attention Profiling Mark-up language working group is working to define a specification for portable attention data, called APML,  that will make our attention data portable.  The goal is to "boil down all forms of Attention Data - including Browser History, OPML, Attention.XML, Email etc - to a portable file format containing a description of ranked user interests".

This is a brand new working group - it is less than a month old - but it includes some interesting members such as representatives from Scouta, BuzzLogic and Digg.  The current spec at draft level 0.2  is quite thin (really just three pages of content).  In particular, the spec doesn't address specific content types (books, music, movies, blogs).   So this is a perfect time to jump in and offer to help, particularly in the music space where data about attention and taste are so very important. (Via the Scouta blog).

Processing is a Java-based  programming language and environment for people who want to program images, animation, and sound. Stephan writes that a new book on Processing has been released by MIT Press called Processing A Programming Handbook for Visual Designers and Artists.  It will soon be joined by Processing: Creative Coding and Computational Art.

I've seen some really outstanding programs written in Processing, including Elias's Music Rainbow.
 

 

 

GrooWe.net is an experiment in collaborative music recommendation. And as far as a Music 2.0 recommender, they don't get much simpler than this.  GrooWe is a wiki, with one page per artist where users (that means *you*) can add and edit music recommendations based upon the artist. [Read More]

Looking for some new and interesting music? Check out the SXSW site. They showcase music from all of the 700+ bands that played at the festival. At the SXSW news blog there's  a 3.1GB torrent containing 739 MP3s of 2007 SXSW bands.

The recent decision of the Copyright Royalty Board to raise rates will devastate internet radio.  One of the biggest potential victims of the new rates is Pandora.  Tim Westergren, CEO of Pandora has been active in getting the CRB decision reversed.  The Download Squad has a video interview with Tim about the recent CRB decision and its impact on internet radio.  It's a good high level overview as to why the CRB decision is bad, and why internet radio is worth saving.

 

Today, I'm listening to my personal iTunes collection.   When I do this I usually I'll pick an artist and shuffle play through the songs.  This does get a little boring.   However, shuffling through an artist is much better than trying to shuffle play my whole collection which tends to bring up my kid's music (high school musical anyone?) more often than I'd like. Today, instead of using shuffle play, I'm using The Filter to give me playlists. The Filter will generate playlists for you based upon a selection of seed songs.  I commented the other day how poorly The Filter seemed to do when recommending music from the whole iTunes store, but so far today it has done a very good job of giving me a decent playlist drawn from my own collection based upon a seed song or two.  

I like the idea of providing tools to help people better explore their personal collections.  Without these tools are iPods are destined become music graveyards - the place music goes to die.  So far, with The Filter, I'm listening to lots of music that is in my personal collection that I have not listened to ever - and I haven't had to risk iPod whiplash that comes with shuffle play.

Jeff Alexander sent me some screenshots today of Snapp Radio running on the Wii. Pretty cool!  Apparently, the Wii has a web browser (Opera-based), that runs Ajaxy web sites with ease. Very nice.

 

There are some new blogs to add to the music technology  blog roll:

• Music Interfaces - a very nifty blog that focuses on interfaces for organizing, browsing, discovering and interacting with music.
  
• MIR Research - blog by Elias Pampalk and Martin Gasser about Music Information Retrieval.  I found out about the aforementioned Music Interfaces blog via this one.
• Medior - An aggregator of various music technology blogs.  This is put together by jherskowitz who also runs muSick in the head - where he writes about music technology.
• Getcha.info - Music business + Music Technology
  

 

 

I gave the new beta release of The Filter a quick try today. The filter is a recommender remora for iTunes.  Select a few songs in iTunes, click a button and get a set of recommendations - you know the routine  by now.   So for my quick test, I tried the song 'Bring me to Life' by Evanescence. Last.fm users call this gothic rock or gothic metal (or even female-fronted metal).   The first two recommendations from The Filter were  Kiss's 'Rock all night' and George Michael's 'Faith.'  I listened to the George Michael song (I already know that I despise Kiss), and I think I would be hard pressed to find a song that was less like 'Bring me to life'.   

Sure, this is just a quick test, but first impressions matter, especially with recommender systems.  I'm guessing that The Filter is just having the usual 'cold start' problem that plague collaborative filtering systems.  On the other hand, perhaps they are trying to create a musical version of the Unsuggester.

Brain Zisk, founder and Technologies Directory of  the Future of Music Coalition has a lengthy interview with John Simson, Executive Directory of SoundExchange about the recent ruling by the Copyright Royalty Board to raise the Internet Broadcasting rates.  The interview is in the current issue of Royalty Week.  Direct link to the PDF is here: March 19 edition of Royalty Week.  Here's a choice quote by Simson:

 I'm tired of the claims that have been made over and over, a sort of arrogance on the webcasters side that we're doing you a favor playing your music.

Today, my goal is to listen to all 102 versions of Stairway to Heaven.  Hey, it's all research into the many shades of music similarity.  It is going to be 5 hours and 52 minutes wondering about that lady who's confused about shiny objects.

Perhaps you remember the Google Image Labeler, a game where you try to guess what labels your partner will apply to an image.  Michael Mandel, (when not dodging NYC bullets) has developed a similar game for music called MajorMiner.   The goal of the game is to label songs with original, yet relevant words and phrases that other players agree with.  

Of course the game is just a thinly disguised method of collecting data about music. But it is fun, addicting, and the data helps the excellent music tech team at LabROSA.

I'm currently number 4 on the leader board. The neat thing is that your score can go up even when you are not playing as other players select tags that you've already applied.

The game itself is well designed. I like especially like how the scoring rules work. Once a label has been applied a couple of times, you can't score any more points for that label.  Without this rule, you could just type 'rock' for each song without even listening and climb the leader board.  The game has a good, minimalistic web design - no Ajax for Ajax sake.  All in all, well done - and it will be interesting to see what kind of data Michael collects,  and  how the MajorMiner tags  compare to tags that people apply at last.fm.

If you are a college student looking for something to do over the summer, consider applying to the Google Summer of Code.  This year Google will pay stipends to  a thousand or more students to work on open source projects.    Of course, here at Sun we are particularly excited to see OpenSolaris and OpenOffice as part of the SoC.  

There are quite a few music-related projects this year at the SoC.  Here are some to consider:

• CLAM ( at the Universitat Pompeu Fabra) - CLAM (C++ Library for Audio and Music) is a project that aims at developing a full-featured application framework for Audio and Music Applications. It offers a conceptual metamodel as well as many different tools for that particular domain. One of its most relevant features is the availability of a visual building dataflow application that allows to develop rapid prototypes without writing code.
  
• MetaBrainz Foundation - The MetaBrainz Foundation  operates the MusicBrainz Project. MusicBrainz is essentially Wikipedia for music: MusicBrainz is attempting to create a comprehensive user contributed music encyclopedia. They would like to catalog all music from all ages, all corners of the world and all ethnic groups. All of their data is freely downloadable and in the Public Domain and Creative Commons licenses.
• Xiph.org Foundation - Xiph.org is an open source project  dedicated to providing open and free-to-implement multimedia technology as a foundation for an interoperable, level playing field on the internet and other digital distribution networks. Over the past 8 years Xiph has hosted development for all the major patent-free audio and video codec development, including the Vorbis, Speex, FLAC and Theora, the Ogg streaming format, and the Icecast streaming media server.
  
• Taste - Taste is an open-source collaborative filtering engine for Java, hosted on SourceForge.
• Mixxx - The goal of the Mixxx project is to build a stable, cross-platform, open source DJ mixing application suitable for amateur and live professional use.
• XMMS2 - XMMS2 is the spiritual successor to the very successful XMMS project.
• Audacious Media Player - Audacious is a successor to Beep Media Player, a GTK2 port of XMMS.
• Rockbox - The Rockbox project started in december 2001 and in it they develop a complete portable music player firmware replacement - including operating system, GUI and application suite. Rockbox runs on a wide range of support platforms including players from Archos, Apple (ipod), iriver, iAudio, Toshiba and SanDisk
• Ardour - Ardour is a 7 year old project to implement a professional quality digital audio workstation for POSIX-like operating systems.
• FFmpeg - Fmpeg is the premiere open source multimedia processing backend library, highly ubiquitous though rarely directly seen. Dozens if not hundreds of open source and proprietary software programs are either known or suspected to incorporate FFmpeg for heavy multimedia lifting.
• IEM - Institute of Electronic Music and Acoustics, Graz  - concentrates mainly in digital signal processing, audio engineering, and psycho acoustics, eg. projects in analysis and syntheses of sound and loudness perception.