Categories
Music

Music Xray

I was recently sent a link to Music Xray. This company claims to have software that can analyse a piece of music, and by comparing the underlying mathematical patterns to those of hit songs, can indicate the probability that the piece of music will also be a hit song.

Maybe the system works, maybe it doesn’t. That doesn’t really concern me. What concerns me is what the very presence of this system means. It’s another symptom of the state of the music industry, an industry which is controlled by people with no interest in music, people who focus on grabbing money in the short-term and are unaware of the damage that they are causing to the future of music.

Music Xray, if it works, will only succeed in homogenising music, stripping out all of the quirks and mistakes and leaving behind a limited set of acceptable songs. And though these songs will be mathematically perfect, they will be devoid of soul and originality. They will cause a chemical reaction in your brain that brings pleasure, but they will not be intellectually stimulating.

Music will not die, because music does not require an industry to survive. However, we will end up with two distinctly different forms of music – the music which has been produced by computers to cause the desired response to make everyone part with their money; and the music which has been written by humans, complete with all its beautiful flaws, potentially even with these flaws exaggerated as a badge of authenticity.

And maybe one day there will be a backlash, and people will tire of being exploited. Then, the industry will reprogram their computers to emulate the flawed underground music, and the “real” musicians will have to prove themselves through different means. In many ways, we have already seen this happen, as the mainstream repeatedly hijacks new movements and then clumsily attempts to mass-produce what they see, leaving a noxious aroma over the rich, promising cultures which they invaded.

Who is to blame here? Is it the music industry for manipulating the market to seize every penny that they can lay their hands on, without any thought for their long-term livelihood? Is it the consumers, for having simple tastes and a willingness to keep paying silly money for the same thing? Is it the artists, for being so desperate for fame that they are willing to sign a contract that will probably never bring them any money? Or is it all of us, for thinking that we are invincible, and that bad things only ever happen to other people?

*UPDATE: I thought that my theory had a slightly familiar ring to it.*

7 replies on “Music Xray”

I find this technology interesting. Its quite new but it appears to be proving itself.

Your immediate reaction that it will ‘homogenise’ music is quite common. It was probably the first thing I thought of when I heard about it. However, I think this is a knee-jerk reaction. The commercial music industry is dying at an increasing rate and they need all the help they can get so its not surprising they would jump at using this tool.

However, I reckon that most people don’t really know what they want or what they like, their tastes change over time and with the influences of their friends, environment and (especially with teenagers) the image they want to portray. So basically, whether a computer likes the music or not doesn’t really a great deal of bearing on what will become popular.

“the music which has been produced by computers to cause the desired response to make everyone part with their money” This is news to me. Sure, computers can produce music but its a long long way from being artistically sound and saleable. People produce music, not machines. The machines just analyse it, and as a music producer I can say with some conviction that many many machines have been used to analyse the music even before it got to the ‘hit song detector’ This is nothing new and has been going on since before ‘pop’ music really existed.

At the end of the day, these hit song detecting computer programs just perform a role which has been previously been filled by only humans, and humans make errors. Elvis Presley and The Beatles were just two acts who were told they would never make it by the human equivalent of these machines and there are probably many many more.

When you consider that humans can have bad days, personal agendas and bias, do these machines still seems as threatening?

Hi, Contayjen.

Thanks very much for stopping by and leaving a comment. You talk a lot of sense.

I’m a computer programmer myself, and I make music on the side, so my perspective will differ from yours in significant ways. I agree that computers don’t make saleable music right now, but they have the potential to do so in the future. If the music can be analysed to the point which its saleability can be measured with near-complete accuracy, then theoretically that same music can be produced by the same program run backwards (speaking figuratively). As the accuracy of these programs improves, we will get closer and closer to the point where the computers are able to make the music. Perhaps we’ll get asymptotically close to that point, but never quite touch it. At this stage, it’s impossible to know.

Elvis and the Beatles were told that they would never make it – they were told by humans. And they soldiered on, because they knew that humans were fallible, and they wanted to make a great big “fuck you” gesture to those that had not had faith in them. Imagine that they had been told that they would never make it – but by an infallible computer. Maybe they would have responded to that by drastically changing their style and trying again, but I think it’s more likely that they would have been completely crushed, and these new genres that they were trying to create would never have existed.

Humans can have bad days and bias. But they can also have good days, where the sun is shining and they had sex before breakfast. A tape is dropped onto their desk, something that they would not normally have given a second listen, but they are feeling adventurous today.

Changes in the music scene are created by human fallibility, just like human beings were created by nature’s fallibility. It’s mutation, and underneath it all, it’s basically just a series of mistakes.

‘Its just a series of mistakes’ haha…I totally agree with that! All the music I make is a mistake, but I use my ears and my emotions to decide which of those mistakes to keep and use.

I don’t however, agree with the notion that just because a computer can analyse something random it can produce something that will be acceptable to our tastes. Analysis and creation are two very different fields I reckon.

For example, a computer could analyse different types of food for food value…levels of salt, sugar, fat, acids, etc, and it might be able to compare a food sample (or in fact any matter) to an existing database of food samples to predict how likely it is to be edible for the general human population. However, this is totally different to creating a food sample from scratch.

Likewise a computer can analyse voice and work out what is being said, but even with a set of artificial vocal chords it can’t put the thought into speaking a poem from scratch. Not yet anyway.

You could bring ins the example of the room full of monkeys writing an famous novel. Sure, given enough monkeys and enough time, this could happen, but we don’t have the computing power yet to generate a big enough room full of virtual monkeys that would be able to randomly arrive at a great work of art. I’m not saying it can’t be done, but comparing analysis to creation is like comparing the technology of skateboards and space shuttles….there is a long way to go yet, so no need to get worried.

I have other reasons to believe in the continued success of ‘human’ music:

Music is written by humans to express emotions, not to emulate them as such. Music often tells a story, and without the trials and tribulations of everyday life to flavour our musical output, even the most ‘talented’ human writer is often let producing second or third rate music never sees the light of day.

Finally, in the music market the way it is right now in rapid decline, I don’t believe anyone with the musical know-how to program such a music-creation system would want to do so, it would be like setting off an nuclear device in your own back yard. So it ain’t gonna happen for a while yet.

Hi,

Why wouldn’t a computer be able to produce something that is acceptable to our tastes?

Computer programs generally don’t work in the “infinite monkeys” fashion that you mentioned above, because it takes an unusably long time in all but the most trivial cases. In some situations it is possible to find a result instantaneously, by just examining the problem and refactoring the variables (for example, find x where x + 5 = 6). If this isn’t possible, an iterative solution is usually used, where you start with a random or non-random test value and then repeatedly improve it. The monkey/typewriter equivalent of this would be to replace the stupid monkeys with smart monkeys, that can look at their manuscript and say “I like this bit… this other bit, not so much.”

You say that comparing analysis to creation is like comparing skateboards to space shuttles, and I get the impression that you expect me to believe it, just because you said it. The fact is, I disagree when it comes to data. Sure, it’s much more difficult to build a car than to look at a car and say “that’s a car”, but that’s because of the physical matter involved. Let’s even out the playing field a bit – consider a person who understands how a car engine works in minute detail (analogous to your music analysis software). Now give them the power to create matter out of thin air. Let them shape it into any shape that they wish, and move it around with the power of their mind. Don’t you think that they’d be able to build a car? Sure, it might take a little bit of trial and error as they refine the sizes of their components, but I think they’d get there reasonably quickly.

Moving on to your closing thoughts, I’d argue that there are a lot of people writing songs about emotions that they don’t genuinely feel, and they are being quite successful with it because it’s got the right marketing package. The current shape of the industry has proven that you don’t need to be sincere to be successful.

Perfect computer-generated music would be the nail in the coffin of the music industry, but only if it were freely available. If the industry could artificially limit the supply of it, using patent law and DRM, then it would have no effect on the status quo. But there is no DRM in existence that has not been cracked, so the only way that they could make this work would be to outlaw computers (or, at least, make it illegal to own a computer that has not been crippled to within an inch of uselessness, and enforce this).

They’d really, really love to do that. Once upon a time, that would have seemed like a ridiculous suggestion.

‘Why wouldn’t a computer be able to produce something that is acceptable to our tastes?’ – Well provided it was programmed to do this, why not indeed? As a computer programmer yourself, you’d probably find it easy to program a computer to create hit songs, right? Of course not. You know I am joking, but you see my point; Its not as easy as it sounds. The expertise required to do that exceeds what any team of people is capable right now. Believe me, there are people trying to do just that but any human with just a couple years experience of playing an instrument can already achieve greater than the most powerful software and its still quite a large leap from there to actually create a hit song.

Looking at your example (find x where x + 5 = 6) you are vastly oversimplifying the equation. But, say for the point of argument that you have the most advanced music jamming software on the market today, and it has a learning ability when connected to our hit song detector. It could, just like a human, create random variations on a pre-determined scheme and test them. Lets just think about that for a minute. There are hundreds if not thousands of real instruments available, and an infinite number of synthesized ones. There are an incalculable variety of melodies and rhythms across all the different genres, as well as differences in tempo, groove, chord progression. And then you get into harmony, which adds an exponential or two onto the whole equation. Combine that with a flow and passion from real human fingers or vocal chords and my head is hurting just trying to think about it. Even the most powerful computer we could conceive would struggle to work it out fast enough to make anyone any money from writing hit songs. The only way would be to thin it down a bit, to give it a bit of help and point it in the right direction, to ask it to make it a bit like X, or a bit like Y. But then you’re in the same position we are now, with artists being told to emulate others.

Moving on to your next point, I feel that you make a mistake bringing humans and science fiction into the equation with your ‘creating cars out of thin air’ argument. Where is the computer involved in this, where is the artificial creativity? There is no argument here.

Regarding fake emotions in music, you are entirely correct. Like artificial additives or poor ingredients in food, most people will lap it up, not caring or even realising that they are consuming junk because it hits the right spots in their senses. Yes, this is a problem, but who are we to tell people what sort of art to appreciate? ‘Fake’ music isn’t as directly harmful and cancer-causing like artificial sweetener or adht-inducing chemical colours. Ideally I’d like everyone to develop a sense of musical taste and make their own decisions about what to buy, but like anything in life, people are pack animals and will just buy what everyone else is buying because that’s what people do. ‘Don’t think about, just go with the flow and you’ll be okay most of the time’ Thats the essence of the sheeple motto. People take recommendations from friends, radio DJs, TV, or anything that has an impact on their lives. How often to people take a chance and buy something they’ve never heard or heard of, when there are probably a million bands or artists on their wishlist?

But, you know what? It hardly matters. Recorded music is already dirt cheap where its not already free like on the internet and mass producing samey hit-like music will only mean that real, soulful music rises in value, and thats all the better for us real musicians. On top of that, people’s tastes are so diverse that no pop hit ever has made it into selling double percentiles. In fact, the most popular songs by the Beatles, and charity singles like ‘Do They Know its Christmas Time?’ only sold to 4-5% of the UK population. So, even with the the best songwriters and the best, most star-studded marketing campaigns on the planet, the chance of any song being good enough to suit the average persons’ taste enough to buy it is miniscule.

However, to assume that this pattern-recognition technology will steer the music industry further into the corporate dark-age is to misunderstand it. There is great potential for it to rescue the music industry from its demise by helping people find just the sort of music that they like. For example, if you ask the average person in the street why they like the sort of music they like, if they aren’t too knowledgeable their answer might be something like ‘I’m not sure, but I know what I like and what I don’t’ In a world where musical borders are increasingly getting blurred and the marketplace is getting evermore flooded with new styles and new bands (there’s over 5 million on myspace alone), having an easy way to find the music you like can only help both fans and artists.

So, this pattern-recognition technology is currently being deployed alongside music databases for use by everyone, not just record labels. Intelligent radio services like Pandora can accurately predict the music their listeners will like and fine-tune it so that you don’t have to sift through hours of garbage to find that diamond. I predict that music digital retailers like iTunes, if not already using it will soon be doing so, to tailor the music marketed at each customer.

Taking a logical step forward, imagine if you had this program controlling playback on your own music database, like your iPod. You’d just tell it what sort of music you were wanting to listen to, and it could instantly compile a stream of music tailored to your mood at that moment. You could have different music for the morning, on the way to work, and other music going home in the evening. Music for waking up to, music for chilling, music for going shopping, music for jogging or gym, even music to get jiggy to, all selected automatically froma vast database by the style, speed, mood etc.

Still think its all doom and gloom?

Hi!
I think there is probably something behind the statement that music can be produced mathematically.
But i dont think that it considers a songwriter/songwriters experience of life itself,
upbringing , family , friends, walks of life , personal drama , cultures , happyness ,
travel , Sadness, etc… witch is were most songwriters find there inspiration.
I personally believe that major music corporations are facing extinction if the don’t
dramatacly change their business venues and think ouside the box. There box being of material concern only and not an combination of business and artistic development.
So no matter how hard they try with all there computer prognosis i think they are lost. Lost there true love for music and that is what it all comes down to. Not to be able to enjoy and appreciate the versatility of music , but to try to pocket it into their own music universe of making money.

Long Live Music!

Leave a Reply

Your email address will not be published. Required fields are marked *