Remastering vs. Reperforming
A North Carolina software company called Zenph created technology that can analyze a recording of a piano performance and then reperform it.
The idea is simple: Old recordings sound old. Decades of amazing musical performances are hidden behind the limits of audio technology at the time they were recorded.
The Zenph “re-performance” process isn’t a remastering — that is, trying to fix an existing recording with equalization or noise reduction. Instead, it’s a new recording of a performance that scientifically matches the earlier one. Zenph uses a Yamaha Disklavier Pro, an actual acoustic piano that can, with a computer’s help, play back with microscopically accurate timing and sensitivity.
The question is: Is this good or bad?
Being a pianist, I can’t decide. I would love to be able to hear higher quality reperformances of legendary pianists like Gyorgy Cziffra and Vladimir Horowitz in their prime, yet using technology in this fashion is the equivalent to capturing a person’s soul in a box to me. If we can breakdown the essence of a musician’s passion and artistry into 1s and 0s, what’s going to stop us from taking it further? Should we?
3 Comments
Sorry, the comment form is closed at this time.
-
Recent Comments
I wouldn’t call it “capturing a person’s soul in a box.”
The software is really just analyzing a pre-existing performance and copying it identically. It’s just a copy. Think of it like this: Imagine you could travel back in time and listen to one of these performances live, and you take with you a modern, high quality recording device. You bring that perfect recording back in time to the present for everyone to hear. Not so bad, right?
The point at which I personally might pause for thought would be when a computer can analyze a given musicians *style*, and then perfectly emulate that musician perfectly with *other* music. Probably not too far off in the future, when you think of it.
Regardless, this current software isn’t scary. I guess I just see it as another way of copying/cleaning up existing, old recordings. And that’s a good thing to me!
I can see that. The part that got to me was the prospect of “fixing” mistakes. It was mentioned in the article that it was possible to adjust the playback. That’s something I hope is never done for authenticity’s sake.
I remember some guy writing software awhile back that could compose music in the style of Mozart. The programmer then published a CD of this music and made money off of it. The debate was, who owns the music? The AI or the programmer?
Ah, gotcha. I missed the part about fixing the “mistakes.” As far as the debate about who owned that Mozart music, the AI or the programmer, I would say the programmer… without question. An AI (in that sense) is not a sentient being with rights like people have. It’s just a computer program. If I right a computer program that creates some sort of art, and I can sell that art for money, then it’s my money and my art. I made the software, so I made the art.