3 ms·

While this is cool, two critical points are (a) why no spectral analysis of K448 to determine what parts of the music are responsible for the effect and (b) IED

While this is cool, two critical points are (a) why no spectral analysis of K448 to determine what parts of the music are responsible for the effect and (b) IEDs are suggestive of the brain region primarily responsible for seizure onset but the jury is still out if they can be used as sole criteria.

gnramires 5y ago
If the emotional connection mentioned in the paper is correct, then the actual structure of the piece would dictate its effect, not a simple harmonic content -- although the Fourier Transform is bijective, so if you took into account details in phase I guess you're right.
Bud 5y ago
I would doubt that a spectral analysis would be revealing, or indeed, of any use whatsoever. After all, such an analysis would vary from performance to performance, or even between recordings of the same performance. It would seem more likely that the musical contents would somehow be the relevant factor, although this is speculative, of course.
spoonjim 5y ago
It’s probably not in the spectrum. It’s probably in those flowing mildly syncopated arpeggios.
hex4def6 5y ago
I guess there's plenty of ways to tease apart the mechanism. I'd start by changing speed, try various frequency shaping (eg, squashing the treble), a version done in MIDI with pure tones rather than wavetable), try midi with wavetable, but with successively bigger changes to the tone, try playing various subsections on repeat, etc.
- est31 5y ago
  Yeah I've sometimes helped debug issues in the Rust compiler. People would come with a piece of code that made it crash or something. One task is to minify the example as much as possible. You start by commenting out as large components as possible, but need to watch out that nothing depends on the commented out code in a way that is detectable at compile time. So you remove piece after piece and finally end with a minimal example that reproduces the issue. In complex projects with lots of dependencies you often start out with large dependency trees and somehow have to reduce it. Sometimes it's easier to reduce something, sometimes it's harder. I guess here it's analogous when one wants to find the "core" part of the music that helps with epilepsy. Maybe one can try to vary it in a way that the effect is even larger.

gnramires 5y ago

If the emotional connection mentioned in the paper is correct, then the actual structure of the piece would dictate its effect, not a simple harmonic content -- although the Fourier Transform is bijective, so if you took into account details in phase I guess you're right.

Bud 5y ago

I would doubt that a spectral analysis would be revealing, or indeed, of any use whatsoever. After all, such an analysis would vary from performance to performance, or even between recordings of the same performance. It would seem more likely that the musical contents would somehow be the relevant factor, although this is speculative, of course.

spoonjim 5y ago

It’s probably not in the spectrum. It’s probably in those flowing mildly syncopated arpeggios.

hex4def6 5y ago

I guess there's plenty of ways to tease apart the mechanism. I'd start by changing speed, try various frequency shaping (eg, squashing the treble), a version done in MIDI with pure tones rather than wavetable), try midi with wavetable, but with successively bigger changes to the tone, try playing various subsections on repeat, etc.

est31 5y ago
Yeah I've sometimes helped debug issues in the Rust compiler. People would come with a piece of code that made it crash or something. One task is to minify the example as much as possible. You start by commenting out as large components as possible, but need to watch out that nothing depends on the commented out code in a way that is detectable at compile time. So you remove piece after piece and finally end with a minimal example that reproduces the issue. In complex projects with lots of dependencies you often start out with large dependency trees and somehow have to reduce it. Sometimes it's easier to reduce something, sometimes it's harder. I guess here it's analogous when one wants to find the "core" part of the music that helps with epilepsy. Maybe one can try to vary it in a way that the effect is even larger.