Category Archives: Music Discovery

#SFMusicTech Summit XII Wrap Up

Metadata, metadata, metadata.  With the switch to digital, everything relies on metadata, and right now the system is broken.  The gatekeepers don’t want to let it go, and the new players haven’t figured a good way around them.  This was the theme going through my head long before the ‘Towards a Credits, Rights & Terms of Use Database’ panel capped the day off.  From Brad Prendergast of Sound Exchange complaining about the state of the data that they receive from webcasters (and subsequently delaying payment of moneys owed to artists), to East Bay Ray wanting a way to track all the content on pirate sites so artists could get paid their fare share of ad revenue,  it kept coming back to a source of high quality metadata that people could freely use.

Here are the highlights of the panels I went to.

Community Conversation: Internet Radio

Kurt Hanson of RAIN, along with Brad Prendergast from SoundExchange led this discussion.  It was interesting to learn more from the people directly playing in this field.  The good news is that there is money to be made through in-stream advertising and banner ads, the bad news – it doesn’t come close to covering the costs paid out to Sound Exchange, let alone support you.  Brad did a good job of deflecting the arrows shot at him regarding rates, and just presenting ‘the facts’.  One of those facts, which I already knew, was that the data that Sound Exchange receives is in terrible shape.  They have their repertoire database which does a good job of decoding this, but this problem is one of the reasons there is so much unclaimed moneys.  If we had a good source of metadata, and everybody used this, that would help…

How Musicians are Making Money

This was where the entertainment really started.  Meant to be a discussion around how musicians can make money, it was soon coopted by ‘East Bay Ray‘ of the Dead Kennedys as a rant against music piracy and Google/You Tube’s low payout rates.  Antony Bruno, the moderator did his best to reign him in, but there was little he could do.  Steve Rennie, the manager of Incubus, did a good job of providing the counterpoint of quit bitching and live in the current world and figure out how to make money.  Poor Kristin Thomson of the Future of Music Coalition was hardly able to get a word in edgewise.

How We Will Experience Music in the Future

This was a standing room only panel where the emotions were running high, all off of one statement made by Max Weisel.  Max made the assertion that someday a computer will be able to create the perfect music for YOU!  Not recommend, but actually create the music, cutting the human being out of the relationship.  This created a firestorm.  Dave Allen did his best to ‘play the adult’ and temper Max, but Max would have nothing of it, and even went so far as to play a bad piece of violin music that was entirely computer generated.

Intellectually I believe that Max is correct, emotionally, I hope he is dead wrong.

Data / Analytics Eating the World

This panel revolved around the how analytics has changed many different industries, from politics to fashion. I found the fashion industry comparisons very interesting, as it is similar to the music industry – based around creativity, and you never quite know what is going to cause the next big hit.  Analytics can help for tracking and predicting the mainstream, but not the outliers.  My take away from this panel – track everything!

Towards a Credits, Rights & Terms of Licensing Database

I was really looking forward to this panel, but in the end, I was a bit disappointed   I guess with all the heated debate shown throughout the conference, I expected a bit more passion – there was some, but nothing compared to last Summit’s metadata panel, where everybody was agreeing that all parties need to work together to solve the problem, but you could see everybody pointing at the other person behind their backs. I heard Zoë Keating saying that we need to figure out how to give the credit so artists could get paid, and she is ‘happy’ to enter her data in order to get there. I heard David King, one of the creators of YouTube’s ContentID system, say that this is a hard problem that costs a lot of money to create and maintain, and once created the data providers don’t want that data in the public, or else Google would have graciously solved this problem for us years ago. I heard Robert Kaye of MusicBrainz talk about the wonders of being a metadata hippie and the life of open, and how if people would just use the MusicBrainz ID…   And then I heard Kevin Lewandowski of Discogs talk about using open data to solve a business problem.

What I didn’t hear was anything from the incumbents – where were the publishers, SoundExchange, PRO’s, the labels. What is going to make their life’s easier if this data becomes freer, or why do they need to keep it locked up.  I didn’t really hear any solutions except that maybe someone should sue the PRO’s to try and get them to release their data.

To me, Discogs is going in the right direction.  There needs to be a business reason to support free data, free data in itself doesn’t quite work.  The challenge becomes how to keep the data free, how to not become one of the incumbents who’s life depends on closed doors, contracts, and NDA’s.  Stay tuned, I will have a post up around this soon.

Wrap Up

I love the SFMusicTech conference – Brian Zisk does a great job of putting on an interesting, affordable show, and most importantly, securing plenty of free beer.  This was a great show and now I’m looking forward to the next one.  If you are in the industry, make sure you don’t miss it.


The Fallacy of ‘Similar Music’

There are many different sources of streaming music available today – Pandora, Spotify, Rdio, etc…  All of these rely on what I call the ‘fallacy of similar music’.  You put in a starting artist or song, and it builds you a playlist of similar music.  While this makes getting started relatively easy, after about a half an hour I get really bored.

As I get more into photography, I am figuring out the difference between a snapshot and a portrait is the ability for the picture to create an emotional reaction in the viewer, music is the same way; a well crafted playlist will make you go ‘oh wow’.

I have yet to find an algorithm that does that.

Do you want to listen to snapshots or portraits?

Do you TRUST your DJ?

So I walked into PlayNetwork the other morning, and was making my morning tea, when I noticed the overhead was playing some classic jazz – no idea who it was, but I was enjoying it.  It kind of struck me that here I was listening to something I was completely unfamiliar with, yet I trusted it was good, and that the next song would probably also be something interesting; I asked myself why I felt this way, and it struck me, it was because I ‘trusted my DJ’.  I know the music programmers at PlayNetwork don’t put together crap, so I’m going to give them the benefit of the doubt instead of ‘changing the station’.

Commercial radio have transformed DJ’s into voices, but don’t trust them to pick music anymore.  What is interesting is that with the tracking capabilities of the ‘Personal People Monitors’ that are used to track ratings, it has been shown how much people will jump from station to station when they here a song they don’t like or are unfamiliar with.  If they actually trusted their DJ, would they be so quick to move?  If commercial radio empowered their DJ’s to be more than voices, and allowed that trust to be rebuilt, would you see as much jumping around?  Looking at the success of stations like KEXP and KCRW, which do have real DJ’s, I think they would – what do you think?

Do you trust your DJ?

Where’s the Spotify Web Application?

So I subscribe to both Spotify and Rdio for on demand music.  Why two services, you ask.  Well, I originally subscribed to Rdio so I could use their API on a project I am working on (more on that later…).  Then Spotify finally came to the U.S., and needing to be one of the cool people, I subscribed; not realizing there was no web application.  I like Spotify’s application, but I don’t want to install it on both of my laptops, the computer in the kitchen, the one at the office, etc.  Also, without a web based app, how do a create an application with their API that is truly useful?  So I guess what I am saying is “hey Spotify, build a web application, or your going to loose me”.  Seems so Rhapsody 2003.

Discovering New Music Through Movie Soundtracks

So I was listening to the ‘Big Lebowski’ soundtrack this morning on Spotify; this wasn’t the actual soundtrack CD but a playlist that someone else had put together, so the tracks were all taken from their original CD’s.  As I am listening to this very eclectic mix of music, it hit me that this could be a great form of music discovery.

What if you could get a giant database of movie soundtrack music and use that for creating a playlists.  Not only do you have the standard relationships between tracks (artist, albums, genres), but you also have the relationships between movies, as well the related emotions attached those movies.  Start with your favorite movie, and branch off from there.  You may discover some new music, and possibly some new movies to watch.

Anybody know of a database of movie music?