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

The Hype Machine is the place to go to find the most fresh music.  Hype Machine crawls the many music blogs and agregates them into a single site that contains all the music that people are talking about.  In the past few years, the Hype Machine has grown in influence to become one of the top-3 destination sites for music discovery.  Let's take a look at the place where the HypeMachine was created.

Here's the grand entrance to the original Hype Machine office:

 

And in we go ...

 

Yep - that's a dorm room.

 The Hype Machine has now grown to a team of 4.  Clearly the office couldn't deal with a 4X expansion, so they've just recently moved... but no pictures of the new office digs yet.

[Read More]

The rainbow room at Spotify - it's in the city, has natural light, nice whiteboards, funky curtains and a tablecloth. This has a nice European feel to the space.

Update: -  Looks like Spotify has outgrown their space. They are getting ready to move.  I hope their new quarters are as nice as the current place. It really looks like a fun place to work.

My co-worker, Jeff points out that all these guys seem to share the same haircut ... perhaps that's a prerequisite of being a developer at Spotify - and obviously that points to a next blog series Haircuts of Music 2.0.- coming to a blog near you.

MusicBrainz provides deep music metadata for Music 2.0 - its founder, Robert Kaye runs MusicBrainz as a non-profit - relying on donations, licensing fees and the kindness of friends and strangers to keep the site going.  Here's MusicBrainz world headquarters:

 Not to say that Robert builds MusicBrainz all alone - there are many, many volunteers that make MusicBrainz happen.  Here's a photo from an MB Summit (notice all of the last.fm guys):

 

And the rack that runs MusicBrainz:

(alright, these are not 'office' pictures, but who can resist photos of a nice rack!)

[Read More]

  
Qloud is a new Music 2.0 company. They have a somewhat distributed work force.  Here's their nifty office.  They have an open office space, work on laptops, with some natural light leaking in.   Since they are a music 2.0 company, I'd expect to see some audio speakers - I'm sure they are there and  just didn't make it into the shot.

 

[Read More]

This one is very nice.

... also bought Durex Condoms.

Here's another example from the Amazon recommender highlighting some of the unusual recommendations that can occur with a collaborative filtering recommender.

Here at Sun Labs there are some folks who think about privacy, security, IP and DRM.  One of these researchers has asked me how DRM has affected my research.  I told her about the problems DRM causes for researchers who are doing content-based analysis of music.  Our music becomes listen-only music that we can't process in any other way.  I'm interested in hearing about other researchers that have encountered problems in their research because of DRM.  So, if you have such as story, send me an email ([email protected]) or leave a comment.  

Here's an excellent piece on the impending death of internet radio, featuring interviews with Bill Goldsmith of Radio Paradise and John Simpson of SoundExchange.

(Via Rusty, Via Mediaor) 

Robert posts on his blog that Last.fm and MusicBrainz have come to an agreement. This can only mean that (hopefully) soon, all of the nifty track, artist and album taste info served up by the last.fm web services will be easily associated with the deep metadata that MusicBrainz provides.  This is just grand! Not just for last.fm and MusicBrainz but for the entire Music 2.0 community.

This happens to me all the time.  I read a blog entry about a new band.  I want to check them out later - to listen to,  perhaps to see if they are listed on eMusic, or perhaps to see if they are touring - but with my 48-year-old neurons - when I finally do have a few minutes, I can't even remember that there was a new band that I wanted to check out.

That's where music.of.interest comes in.  music.of.interest gives you a tiny bookmarklet to put in your bookmark toolbar.  When you click on the bookmarklet, it brings up a tiny window where you can type in the name of the band that you want to remember for later.

 

This just adds the artist to your list of music.of.interest artists.  The next time you are ready for some music exploration, you can just pop over to the music.of.interest site.  There you'll see your list of artists, along with lots of tools for exploring these artists.

With m.o.i you can easily look up info about the artist at the usual places like Wikipedia, AllMusic, Last.fm, Flickr, eMusic, the hype machine and Pandora.  You can tag bands, add them to your eMusic 'To Buy' list.

music.of.interest - is quite fresh - I could still smell the paint drying.  The community of users has grown to about 12 right now, and there are a few unimplemented features.  The main missing feature is that you can't listen to any music on the site.  You can read bios, find videos and photos, but if you are exploring for music, you really need to be able to listen to the band too.  Craig Huizenga, author of m.o.i says in his blog: Hopefully soon you will start to see artist and album information on the site - I’m going to source the base data from the excellent Music Brainz  database, and then try and tie that over to the Rhapsody.  After that, it’s tracks and playlists ...

m.o.i - is a well designed, clean site - written in Rails - it doesn't overwhelm with ajax, intuitive to use ... but is clearly still in development - there are planned features that are just not implemented yet. 

I've added the m.o.i bookmarklet to the very precious real-estate that is my bookmark toolbar (right next to the javadoc for the jdk).   I'm looking forward to seeing all of the things Craig is planning to do with the site.

I've been involved in the JCP expert group defining the next revision of JSAPI 2.0 for a long time.  The expert group has been meeting weekly for years! It's not because we are particularly slow, or lack motivation, it is because speech is hard and defining a speech API that is complete and easy to use is very hard.  The years of work are finally coming to fruition.  The expert group has released the proposed final draft.  Check it out on the JSAPI page.
 

Here's the tee-shirt example. (thanks, steve)

The notifications about which papers were accepted to this years ISMIR were emailed on Friday.  Of the 214 submissions received, 129 were accepted (34 as long papers, 28 as short papers and 67 as posters).  The list of accepted submissions are already on the ISMIR Programme site. Some of the papers  that I'm looking forward to reading are:

•  A semantic space for music derived from social tags by Levy and Sandler - looks to be related to some of the work we've been doing here at Sun
• The Quest for Ground Truth in Musical Artist Tagging in the Social Web Era by Geleijnse, Schedl and Knees - a nice title (a nod to  the oft cited music similarity ground truth paper), and a difficult topic.
  
• Identifying words that are musically meaningful - Torres, Turnbull, Barrington, Lanckriet - I really enjoyed Doug's paper from last year - this paper looks to extend that work.
• MusicSun: A New Approach to Artist Recommendation -  Pampalk, Goto - Elias and Masataka are well known for their music interfaces. And the paper has 'Sun' in the title.
  
• Music Recommendation Mapping - Donaldson - Justin has done some really nice visualizations of the MyStrands data in the past - looking for more here.
• Classifying Music Audio with Timbral and Chroma Features - Ellis - curious to hear what Dan has to say on this topic

In addition to the regular papers - there are 4 3-hour tutorials scheduled for the event:

• Synchronization and matching techniques for music data, Müller and Dannenberg
• Music Recommendation - Celma, Lamere
• Introduction to the MIRToolbox  Lartillot
• Techniques for implementing the generative theory of tonal music - Hirata, Tojo Hamanaka

I had 3 submissions accepted this year - the Music Recommendation tutorial with Oscar, a 2-pager demo paper describing some of the visualizations in SITM, and a poster that is describing some of the work that Doug, Thierry and I have been doing over the last 6 months on predicting social tags. 

I am really looking forward to the conference - the programme looks to be excellent and the venue in Vienna will be a great place to visit in September.

Sun is a traditional campus environment.  We don't have cubicles, we have offices with doors.  If you are of a certain level you get an office with a window.  There's plenty of 'team space' .  Here's a typical Sun office (this is Romain Guy's).

Of course, not all Sun workspaces look so spartan.  At the other end of the spectrum is the work space for Arshan.  Arshan works in Sun Labs on the Spots team. Arshan has a knack for acquiring cool things for his workspace. 

 

Dual 24" displays, a sound system that could fill a barn in a lab that is filled with other high energy researchers and engineers.  A very fun, dynamic workspace (albeit a bit chaotic).

By comparison, our shared space in the east coast is incredibly boring. Here's a shot of the SITM team.  We are working together on a paper, so we camped out in this conference room for a few days.  Very productive, but very temporary.  After this week, we'll all go back to our own offices to sit in the dark alone.

Zoomquilt2