Changelog

I went to a pretty interesting talk this morning by Jen Mankoff about handling ambiguity and errors in recognition systems. Though the coolest thing wasn’t really related, but was a prototyping tool called SILK. It lets you sketch out a design on a computer, then renders that design into interface components. This is really cool! The problem that my UI group has run into is translating our paper designs into a testable interface without having to use something as high-fidelity as Java or VB. I really want to try this now.

update: So I was pointed to DENIM, a related project put out by the Group for User Interface Research at Berkeley.

update: James Landay is the person responsible for both SILK and DENIM.

Now it seems to me that digital music inherently invalidates the whole idea of the “album,” in so far as there is no longer a physical limit to the amount of music distributed by an artist (as there is on CD/Tape/LP/etc…). However, it also seems to me that the “album” is also representative of a collection of music that is in some way related. I will therefor use the term “collection” to refer to this body of work (album carries to many connotations).

So… With the definition of “collection” in mind, I propose that we need a means of distributing a collection of music as a corpus, rather than as a collection of separate pieces, today this is accomplished by distributing a single large mp3 (those who listen to live music know what I mean), but this has the drawback of not being able to provide information on the individual pieces contained within the collection. Now, it may be that the artist would rather that information NOT be available, and for them the very-big-file method works just fine. But on those occasions where it would be nice to provide information to the listener about individual components of the collection, we need a new system (suppose you want to pass the Beatles’ White Album on to posterity as you listened to it on LP).

To this end I would propose the “collection” file format, of which I have conceived two different implementations:

1. A single file in which each individual file is stored, separated by some distinct marker, prepended by a header containing pertinent information about the collection and the files it contains.
2. An archive consisting of the constituent files and an xml information file detailing information about the collection.

Well, my web page is back… Again… This seems like a much easier design to deal with.