Interview with Matthew Dunn, CEO of MusicIP

MusicIP is the killer music search engine. When I say “music search” I mean finding things that are musically similar to other things. Not words, music.

For listeners of music we solve two problems. We call them recovery and discovery. The recovery problem with digital music is that if you like it, you get too much of it and it becomes unusable. Average collections are 500 tracks now, but it is not atypical to talk to someone who has 5,000, 10,000, or 15,000 music tracks in their library or even in their pocket. You can’t remember that many different songs and that many different artists.

That’s right, and the second favorite is the built in accompaniment to the shuffle, the skip. Random music selection doesn’t actually suit what people want out of music. They really want music to fit their moods, as a background for life.

The solution we bring works like this. Find one song that you like. Find one song that suits the mood right now and we will find all the others in this collection that are musically similar to that.

That’s the secret sauce and six years worth of research and development. Our fundamental approach is a balance of analysis of the acoustics the actual sounds in the track, and layering on of social and popularity data to fine-tune the results. We take a piece of music once, somewhere in the world, and put it through a very intensive algorithmic analysis. We boil down the musical characteristics of the track in a way that we can relate to any other track. The punch line is that we’ve done that with just about 23 million tracks to date.

Yes, it’s more user centric, more “user in the moment”. I may like loud banging rock ‘n’ roll when I am driving home from work, but I may feel like Mozart on Sunday morning. In both cases, I have 10,000 tracks to choose from and I really don’t feel like thumbing through 10,000 items to make those choices. I’d much rather listen, enjoy, and get on with the other stuff that I am doing.

Well for today we provide software that makes the music that you have instantly accessible. That’s the recovery problem I mentioned earlier. Today we also point to additional music in the world that is similar to what you have or what you are listening to. We don’t have a service on the market yet where we can actually play that music for you. That’s what we are working on like crazy, as we speak.

Correct. There are a lot of ways to discover music. You mentioned one of the primary ways, which is social. My friend who listens to a lot of music said I might like X. There is really no substitute for that, but it’s not quite as potent as actually hearing something and deciding whether you like it yourself.

I joked with someone the other day. I said there are three ways you get music; you’re sold it (it comes on a CD with the one you really wanted), you’re told about it by someone you trust, or you actually hold it in your hands or your head for a minute and say, “Now that I really like!”

The enterprise business is a bigger piece of our revenue at this time because we have elected to take the desktop software implementation and release that primarily for free. The value that brings to the company is that that’s where we get access to music and that’s where we get CPU time for algorithmic analysis. Consumers are our extended music analysis farm, which in turn helps us provide a better solution to those music enterprises that are trying to get back to those same consumers.

Well the music enterprise has a 100x version of the recovery problem. If you’ve got a recovery problem, how do I pick what I want to listen to in my library? The enterprise is trying to sell it. The typical music enterprise has around two million tracks as part of their offering. No consumer wants to sort through two million things to pick what is fundamentally a mood product. Our offering for those enterprises is that we can take the catalog and wrap it around the customer’s taste. This is done with surprisingly little work because the base technology is built on music, not metadata and not other extraneous nonmusical information.

There is a third leg of the stool that you actually have not mentioned because we have not made a lot of noise about it yet, but this same magic of relating pieces of music that we can do on a desktop computer or in a commercial collection of millions we can actually do in firmware on a device as dumb as an MP3 player or a mobile phone. The first licensee of our mobile technology is a company will have three new MP3 players on the market this fall. All three of them have MusicIP technology built into the one button play list. It’s right here if you look at the button.

To answer that one would take a longer conversation. We find fundamentally that we have to identify enterprises who think they are not doing as much as they could. Who don’t think they are realizing the potential of their catalog or of their device and who are looking for something that changes the experience with their consumers in a fundamental way. One of my colleagues said it’s like trying to have a religious discussion with them to get them to consider whether or not we have a better solution. Of course we all believe that we do. You look at the constraint of the devices and the on-line music stores that in the market right now and it’s not particularly pretty. The average IPod isn’t anywhere near full. In fact, it’s mostly empty, and obviously that is not because of lack of storage space. It’s really because you can’t use everything you could put on there. The boundary is what a person can do with that. We can change that.

Similarly, a service with millions of tracks to sell typically is starting to sell down into the more obscure portions of the catalogue, but most of them have told us they are really not there yet. In fact, I heard the head of the editorial department at Rhapsody (which is a big music service) speak at the “Future of Music” conference last week where I was on a panel. He said their editorial team only has time to help expose 10 or 12 percent of what they have, which means 88 percent of the stuff is left for consumers to patiently dig through and try and taste match. That’s the equation we think we can fundamentally alter.

You noted that we have been in business since 2000. Most of that, I guess by good fortune was very intensive R & D on this very hard problem. We’ve actively been on the street as a business since we were spun out as a stand-alone company in May of 2005.

Nine months later, we rebranded ourselves as MusicIP. We’ve been at this for a year, not quite a year and a half trying to see if there’s business in licensing this technology.

Three really stubborn guys. One Dr. Adam Kolowa who’s the founder and CEO of a company called Parasoft and two employees, two of his top folks who he put on this project that they all believed in, which was we can get to the stage where algorithms reveal enough about music to make meaningful relationships between tracks. Parasoft incubated that company within a company and incubated that research for just over five years, that’s a real leap of faith in applications.

I prefer to think of it as massive good fortune on my part because I can’t think of anything more engaging than this. It is like the most phenomenal thing that I’ve ever been involved in. But the narrative part of that is I was brought in by the Venture Capital Firm that was looking at an investment in the company. They brought me in to do what’s called diligence finding out if the technology’s for real. I happened to have an unusual background- both a technologist and a musician, so I guess I was a pretty good fit for that. I frankly quite quickly determined that the technology was not only real, but was world-class and more or less made a shot in helping that company realize that potential.

Well, the sales pitch I’ve missed a couple of times if we’re working on a consumer facing service, we felt that our sales pitch was coming up short because we couldn’t point to a compelling large scale experience and the underlying business behavior purchases, listens and things like that that said that this in fact a better way to do it.

So in absence of getting someone else to do that, we’re working to launch a music search engine based on our technology and partnerships within the music industry as fast as we can. We think that that’s going to give us the compelling example that’ll help us on the enterprise licensing side. Because the odd thing about digital music is because the business model is getting extended as we’re talking and because it’s built on top of a very strange legacy of 100 years of rights and licenses and legal instructions that don’t necessarily make a lot of sense for a digital era.

There’s not much margin left in selling digital music. Plain simple fact, there’s just not a lot there for companies in that business to work with. So when they’re approached by salesmen or technology vendors, who I think they probably see as co-equal, their natural response is, “Uhhh, I’d really love to, it sounds really interesting, but I’m dealing with a couple of cents left over from every purchase, can you really prove to me that you’re really going to make enough of a difference that I should contemplate taking the plunge with you?”

Yeah, I think so for the moment. Music was a pretty small industry when our parents were buying 45′s. The marketing sales goals were to sell a single. Netted on the sale less than a buck. Nobody had invented the LP. The industry invented the LP and led to a fairly prosperous period of almost 50 years where the single you heard on the radio or the single your friend told you about sold on a 10 or 12 or 15 dollar LP, or CD, which is just a digital LP.

Now we’ve gone backwards in a sense and we’re back to singles and it’s very difficult, I think, to make that work. Is it going to get restructured or are people going to start going buying collections of some sort, is the album going to survive, is it going to be in a compilation, is it going to be there? We don’t know, we don’t know the answers to it, but we know that one of the chief pushes for digital music in terms of the content side of the industry is making singles a really viable business.

Ah, I think our search engine is a pretty intentional comparison because I may not like the fact that if I go searching for Windows that the third entry in a search engine is it, they have a java engine for Windows, just throwing out examples, but the search engine stance is kind of “look for that.” Our credibility and our trust from consumers comes from us doing our best to be relevant and objective in a way the technology underneath it works. That’s our position as well.

The guy whom we stand to benefit disproportionately for the short term is probably a talented, under funded, independent musician who doesn’t have the marketing budget to crack through and make a hit, which is the record industry hold, but they have something that’s really appealing to a lot of people out there if they were just exposed to it.

You know, we’re not going to change the search engine stance and introduce aiming of our own technology and try and skew towards one type of music over another based on label relationships. In fact, for one thing, we can’t, it actually doesn’t work that way and I don’t think we could make it do that if we tried.

With our system the consumers are in charge. The average 19 year old does not buy albums, they buy songs that they like, they buy tracks that they like, and they accumulate a lot of them. Is it then better for the few remaining big labels that get kids capable of cherry picking?

Their stance seems to be, “No, no we don’t like that because we’re getting a one dollar sale instead of a 15 dollar sale.” The consumer stance is,” I like it because I buy what I want.” The historical lesson is this: people will end up buying more if the roadblocks are taken out of the way. They’ll end up buying more. Historically, that’s always what’s happened with the new more flexible vehicle or media, be it in music or film or whatever.

Fundamentally for me is the base that we’re working with. Like I said earlier, I really couldn’t imagine anything as engaging. Why is that, I don’t know, because I’ve been in technology for 20 years and the arts for 30 years and now I’m getting the opportunity to make a meaningful contribution to both fields at the same time. How much better does it get than that? It’s certainly been an uphill battle. We’ve been incredibly fortunate in having not just funding, but day-to-day involvement and backing from our investors. These guys are great partners.

Even with their backing it’s a hairball to try and make a breakthrough technology business go somewhere and stick. And you same the same things over and over and over again and one out of X number of times someone says, “I get it, I like it. I’ll take a chance,” or, “I’ll expand my thinking a little bit.”

If you don’t have a fundamental “can’t walk away from it” passion for what you’re going to do, you’re probably going to have a hard time making it unless you’re lucky and even if you do, you’ll be tired. I’m never tired of this, I wish there were more hours in the day so that we could do this better and do it faster and do more of it.

The second thing is everyone fancies that the industry that they work in is the hardest and the most complex. Fair enough. I certainly see the emerging digital music thing as a total hairball and sometimes I want to bang my head on a wall trying to figure out how to make our business a part of it and how to make our business a profitable part of it. The state that an entrepreneur walks into, if they really are an entrepreneur is going to have the same characteristic, so sign up for some frustration and have a vision that’s going to carry you through and past that frustration because you know you can do something better, you know you can add value to it.