Long time no see guys! Since it has been 2 weeks from my previous post, today’s update will be a bit longer than usual.
As I mentioned in my previous blog post, I conducted a virtual “interview” with Emeritus Professor Roger B. Dannenberg of Carnegie Mellon University. Thus, I will dedicate most of this post to what we discussed.
———Beginning of Interview———
Me – Why is AI helpful for musical growth?
Dr. Dannenberg – In some sense, anything that isn’t routine with computers is called “AI”. When I type text messages, my phone guesses and pops up completed words to select. This would have been called “AI” in the past. I’m not so sure now because it seems like a routine task. I think face recognition and other “AI” will soon be so routine (and so shallow in terms of human intelligence) that we’ll just call it “automation” or “computing.”=-3AI is helpful because we’re really talking about extending the kinds of problems computers can solve or help with. Music involves a lot of creativity and intelligence, so “AI” is needed if computers are going to help with music creation and processing.
Me – I was intrigued by your lecture on the software you created years ago that accompanies your trumpet playing. What was the idea behind creating it? Would you say the creating process was harder back then, compared to your more modern projects such as Audacity?
Dr. Dannenberg – At the time, computer music was mostly non-real time, and it was common to create computer music on tape and include live performers that had to learn the tape part and synchronize to it. This seemed very computer-centered and unmusical. Digital synthesizers were just getting to the point of real-time synthesis, so I thought here was a chance for the computers to follow the humans instead of vice versa.
Yes, development was a lot harder because computers were much slower. Of course, we had simpler systems. E.g. to put an “A” on the screen, you had a small bit pattern, maybe 7×12 pixels or about 12 bytes that you would write directly to video display memory. Now, an “A” comes from an anti-aliased font library and getting an “A” on the screen requires a whole GUI framework, and users expect to have support to change font sizes and work with different screen resolutions — in some cases, it’s much more work, and programs are certainly much bigger than ever.
Me – What would you say is the main benefit of AI music software and hardware? Is it to make music more accessible or something else?
Dr. Dannenberg – AI and computing help us explore new directions in music. I’m not sure it’s any different than, say, the invention of the piano, which had a profound impact on classical music.
Me – Do you think that there would be a point where the main purpose of AI music shifts from its current benefits? Would that create harmful effects to human musicians, such as making them obsolete? If so, how could the balance between musicians and AI music be maintained?
Dr. Dannenberg – I think audio recording was the big revolutionary technology — it nearly destroyed live music and also created a new musical economy and culture. I think the impact of AI will be smaller but of course some people will write music with AI instead of hiring composers/performers, and some will be employed to develop and use technologies enabled by AI, so there will be some shifting.
Me – Would you say future AI music software/hardware would make more people get into the music industry, assuming it makes getting involved with music easier?
Dr. Dannenberg – I think music will be more of a pastime and less of a profession — similar to the recording revolution but on a smaller scale of change.
Me – On the other hand, would that bring detrimental effects, such as where people won’t learn instruments anymore?
Dr. Dannenberg – I think *more* people will study music, many with AI tutors.
Me – There has been some considerable debate on copyright issues associated with music created by AI software, such as whether it belongs to the software itself, its owner, or someone else. May I ask what is your standing/opinion on this, if you have any?
Dr. Dannenberg – First of all, copyright has been terrible and should be overhauled. Granting control over works for 90 years or whatever is not good for society. I don’t think we’re anywhere close to wanting to grant personhood or property rights to machines or software, so rights will be held by people. People who create composing software can sell it with licenses to control, for example, whether generated compositions become the property of the user or of the original software creator. Similar things happen now with, for example, clip art, where you can get an exclusive license, a non-exclusive license, a free-for-personal-use license, or a license to distribute N copies. I don’t think copyright law has this flexibility already.
———End of Interview———
As you can see, this is a lot of information and I have learned so much! One thing that especially intrigued me after our discussion is the subject of copyright, which is much more complex than I thought. I’m currently finding and reading academic sources regarding the issue of copyright of computer-created music. I have also started to put together my paper and presentation for May 22nd.
That’s a wrap! Thank you for reading as always, and I will see you in my next blog post! Take care, Peter Li
Sources used in this post: Dr. Roger B. Dannenberg, Carnegie Mellon University