I've received a few emails over the past couple of weeks asking when Band Metrics is going to open up its beta, and since there was a discussion in Pho yesterday as well, I thought I'd provide a public update. First, many, many thanks for everyone's interest, support and patience while we continue the development of our architecture, as we're working hard to solve a few interesting, but complex academic problems relative to music information retrieval. And, we're just a small group of guys with limited resources trying to build a machine learning framework with natural language processing which will allow artists and music industry professionals to eventually do some really cool stuff like trend measurement, sentiment analysis, reputation management, and maybe one day scratching the surface of the behaviors relative to music. So, needless to say, we're jammed, but we're getting close to releasing our initial offering, and hope to release our initial beta to the thousands of people now waiting patiently to use our system within a few weeks. But, please keep in mind that achieving these goals is difficult, and it's going to take some time to get there, so what we provide with our initial beta is just a fraction of where we are going and what we will eventually offer, as it's a process that's going to require a community effort.

With that said, one of the questions raised in the Pho email list was why Band Metrics was only tracking 1.3 million songs compared to the 7 million artists Band Metrics is analyzing... Great question!  The answer is that our data mapping engine is running into some conflicts deciphering some of the misspellings and artist variables being pulled by our information retrieval and agent framework. For example, a song written by Eric Clapton may have been recorded and released by Cream, Blind Faith, The Yardbirds, as well as Derek and the Dominos, so our system has to automatically distinguish which version of the song (live, studio, acoustic, etc.) is associated with which band, year, release, etc. It's not a trivial problem, but it's solvable, and as soon as we resolve these issues, hopefully in a couple of weeks, the song number will increase by more than a factor of 30, as we already have the songs in our database.

So again, many thanks for your patience and understanding. In the meantime, here is a real sample of some of the data we are collecting about artists from around the world:

this image shows the breakdown of fans for a band across several social media places (we pull from many more sites than the ones shown) 



this image shows songs plays across various sites for the band, A Times Beach Crush Factor

Btw, I would have responded via Pho as well, but I canceled my subscription a long time ago, but I've got a new subscription request in, and I'll try to respond to future discussions there as well.