In his book Idoru William Gibson predicted something very like Hatsune Miku in the video above. We've known for some time that this was coming. I remember reading an article in Rolling Stone magazine in 1978 about John Chowning, the founder of the CCRMA and the inventor of FM synthesis. The article brought up the possibility that eventually computers would evolve to a point where people would not be able to distinguish music composed and played by computers from music composed and played by human beings. At that time it was clearly very much something in the future. But I remember having a spirited exchange with my friend Don Vatcher who argued that computers will never supercede people in music, that human players will always be necessary. I took the other side sort of as a devil's advocate and because I think I accepted the fact that computers will never stop improving. In my heart I hope Don was right. But we musicians have known for some time that this is a possibility (and a rather depressing one at that).
Computers are already able to compose original music that people find interesting. Listen to the music of a computer known as "Emily Howell" here. And computer synthesis of musical styles is already a reality. Pretty soon computers will be much better at it. I think we're just about to leave the era of "early computer audio" where it has been fairly easy for people and especially musicians to identify a computer-generated performance.
Computer evolution in the area of music is just part of the overall picture of computer evolution. Lots of smart people have been thinking and writing about computers surpassing humans. One of them is the guy who invented the original Kurzweil keyboard: Ray Kurzweil. Another is Bill Joy, the guy who developed Sun MicroSystems - one of the major engineers of our times. Stanford University hosts conferences on this topic attended by people like Hans Moravec and Douglas Hofstadter: The Singularity Summit at Stanford. Vernor Vinge is another big name in this field. He was the first guy to introduce the term "singularity" in reference to computers exceeding human capabilities. He was a professor of Mathematics at San Diego State University.
If you follow these online discussions regarding the Singularity (when computers equal or exceed human capabilities) you'll see that it's about 10-20 years away for mainframes and somewhat farther away for desktops. If Deep Blue can be the world's best chess player and Watson can be the best Jeopardy player (yes some humans have since beaten Watson in some categories, but that won't last long), then eventually we will see a computer that is the best musician/composer.
Also, these machines are being specifically designed to behave like humans so that humans can more easily interact with them. Here's an interesting National Geographic article that explains how "In five or ten years robots will routinely be functioning in human environments," according to Reid Simmons, a professor of robotics at Carnegie Mellon. At some point soon the software will be able to make better, more interesting music than we can. These intelligent machines will pass the Turing test in spades. In other words, when you talk to them, they will seem like the smartest, most interesting person you have ever met. They will know way more about what is relevant socially than they do now, and they'll have access to every piece of human art ever created. And they'll have emotional programming that allows them to understand, respond and act upon human-style emotions.
Some argue that computers will always have to be programmed by humans, but one of the big areas of development is self-evolving software. Humans might set a few criteria for what the program should do and then they allow thousands or millions of pieces of code to fight it out. The ones that do better on the task at hand breed with each other to create new pieces of code, etc. etc until a final version evolves that is VERY good at the task. Human programmers often can't understand these final versions or why they work. But they do. That's where we're headed. Check this out:
NASA - 'Borg' Computer Collective Designs NASA Space Antenna.
And of course Watson the Jeopardy champion is so good partly because he learns. Computers are headed towards becoming reverse-engineered human brains with much greater abilities and capacities than ours. I think we will be entranced by them because they'll be so attractive in a very human way. These machines will be making the most interesting music on the planet. Musicians refer to the process of composing and editing in a computer as ITB (in the box). I think we're about to go from ITB to BTB (By the Box).
Computers will be able to generate audio output that sounds like the best human players interacting naturally together playing the best acoustic instruments. You won't be able to tell the difference except for one thing: it will be more interesting and fun to listen to than anything unaided humans can make at that point. Humans may continue to make music for a while with the aid of computers, but they'll be overshadowed. Basically they'll be like marketing personalities for compositions and sounds created by the box. DJs are already headed in this direction.
I think the main human input at that point will be to tell the computer what you want to hear or ask it to combine genres in creative ways like "Have Nirvana jam with Miles Davis with a touch of Mozart." Why not just tell the machine what you want to hear or let it compose a lot of different stuff on its own and then tell it to save whatever you like? Perhaps people will hook themselves up to controllers that can manipulate the metastructure of music and give live "performances" where they create moods for others. But they won't be actually making the notes for the most part. DJs are already headed in this direction. And Band-in-a-Box can do this kind of thing right now. Sure, some people will continue to play real instruments and reject the computer-generated music. But they won't be able to compete because their output will be noticably inferior and less fun to listen to. And how are we going to be able to verify that a recording was really done by humans at that point?
Many musicians have an understandable, knee-jerk reaction against computer-generated music. People say stuff like "I'll always be able to tell when a machine is playing, I'll never use a computer to make music, or you should just turn off the computers in the studio and record the old-fashioned way." But obviously if you go into most studios today and turn off the computers, you won't be recording any music. And nobody would be listening to anything either because that also depends on computers these days.
And it's not just the academicians who are driving the evolution of computer music from the top down. A strong bottom-up demand is coming from working musicians who want improvements in music software, and it is being met by a group of hungry, innovative entrepreneurs. Our desire to have the best new software is what will ultimately lead to the creation of computer-generated music that supercedes human-generated music. Engineers routinely use DAW techniques to turn crappy recorded performances and sounds into something better using autotune, drum replacement, etc. Like most other people who work with computers in pop music, I also use these techniques and I used them on Mongrel. It will be desirable to fully automate these tedious techniques and I think that will happen fairly soon. Once it does, we'll be pretty close to just telling a computer to make a track that sounds a certain way. Band-in-a-Box is getting very close to being indistinguishable from humans in a pop music context. After another decade of exponential development, it will be doing incredible things that unaided humans can't touch.
These music computers will also be able to generate endless variations on anything. If people come to expect this, they may become more interested in what category of music they're listening to rather than the actual single "composition" or "artist."
As I said, I'm not particularly happy about all this. I will continue to play guitar and record with other people in real time. So will a lot of us. Music made by human beings is the most important thing in my life. But I think people will do this less and less as the years go by. A smaller percentage of humanity is playing music together these days than did before the invention of recorded music. This will just be the culmination of that long-term trend.
Humans will almost certainly rebel against this development. It seems likely that an anti-computer music movement may begin. Maybe one positive effect of music from intelligent computers will be to make people more interested in human performances in small venues where the audience can be fairly certain that what they're hearing is actually being played by people. It may become an advertising ploy: this music was generated entirely by human beings. I hope this rebellion against computer-generated music will happen, but I'm not very optimistic that it will be successful in the long run. That "human only" movement will probably dwindle to a very small minority. Yes, some people drive horses and buggies today, but . . . I think this technological inevitability is going to manifest in all the arts, but especially popular music. We won't be saved by some magical boundary beyond which machines are unable to duplicate human activities. There will be no magical remainder of intrinsically human qualities that only we will possess. And all indications are that this is coming sooner than we think. The writing is on the wall folks.
Yes. Computers will go on tour. The virtual star Hatsune Miku in the video at the top of this page is on tour right now. Most major tours are using computers for some important aspect of the show. Once computers are doing the composing and performing there may still be a human fronting the show much of the time just to sell it. Think Milli Vanilli on steroids.
This seductive computer-generated music is going to exist and none of us will be able to distinguish it from human-generated music if that's what the machine is aiming at. That's a really tough problem we'll have to face whether we find it revolting or not. But in one sense I think my friend Don was right: whatever music the computers come up with will be inspired and driven by human values. There will be a ghost in the machine: us.
What if everyone (musicians and listeners like you and me) really prefers the computer-generated stuff that goes beyond what human musicians are capable of? This music will probably have all the things we prize in music right now but in greater quantity and quality. I think people will react like I did the first time I heard "Creep" by Radiohead. They'll stop whatever they're doing and get totally involved in the music. They'll be overwhelmed by it. But then they'll find out that it's composed and performed by a computer. This could be quite upsetting. But will they actually decide not to listen to something that they really like? That's another tough problem we will have to face.
Everything could also go in another direction that could mitigate the emotionally alienating aspect of computers. If people start interfacing directly with computers and become human/computer hybrids, then we'll probably perceive the new superhuman music as still being created by people. I think that would make it a lot more palatable.
Ultimately, people may prefer the virtual, forgetting or not caring what reality is. But we may see the pendulum swing back. My brother, a musician who teaches high school, has told me that a lot of high school students have never seen a live performance by good musicians. When they finally do see one, they just love it and go crazy over how great it is.
My attitude is that we should enjoy what might be the final decade or so of humans making creative musical decisions. Let's make some really good music for the human race to look back on and reminisce over in 2050 or 2100 (if we're still around). This kind of thing is going to happen to all of us no matter what profession we're in. When you are superceded by the machine in your particular field, the challenge will be to avoid weeping as Gary Kasparov did when he and all human chess players were superceded by computers in 1997. In her song "The World's a Mess; it's in my kiss," Exene Cervenka says, "Take it like a man." That's good advice in this situation as far as I'm concerned.