You folks might be interested in this discussion of distortion on the SS board. I was postulating that we may be looking at the topic of distortion in the wrong way and that may explain why we have trouble correlating measurements to sound quality.
Are we looking at it wrong
Are we looking at it wrong
Lots of study.
I think the answer is a lot of experimentation.
It would take a great deal of subjective testing to get a handle on what sonically intrusive distortion looks like. There would probably be many different patterns that fit this catagory. Then the big task would be to correlate measurable parameters to these shapes. Things to investigate would be areas between the input and output curves, angles of deviation, d/dt of the input and output signals etc. At this point I have no idea what the correlations might be but I suspect that it will be a differential equation of some sort that gives us the answers since the ear is "measuring" pressure changes. Maybe dI(t)/dt - dO(t)/dt (where I(t) = input signal and O(t) = output signal would be instructive.
Of course I may be full of hooey and maybe there is no mathimatical representation that correlates to ugly sound at all.
I could believe that mathematics might not be able to model everything in creation but my guess is that it is possible to get close if one is clever enough. Alas I am probably not. 
I think the answer is a lot of experimentation.
It would take a great deal of subjective testing to get a handle on what sonically intrusive distortion looks like. There would probably be many different patterns that fit this catagory. Then the big task would be to correlate measurable parameters to these shapes. Things to investigate would be areas between the input and output curves, angles of deviation, d/dt of the input and output signals etc. At this point I have no idea what the correlations might be but I suspect that it will be a differential equation of some sort that gives us the answers since the ear is "measuring" pressure changes. Maybe dI(t)/dt - dO(t)/dt (where I(t) = input signal and O(t) = output signal would be instructive.
Of course I may be full of hooey and maybe there is no mathimatical representation that correlates to ugly sound at all.
I could believe that mathematics might not be able to model everything in creation but my guess is that it is possible to get close if one is clever enough. Alas I am probably not.
Re: Lots of study.
http://www.gedlee.com/distortion_perception.htm
!!!!!!!!!!
Look at the second presentation on this page!
I've posted this in several of the distortion threads, but no one is commenting on it. Maybe no one is looking at it?
It may not be *the* answer, but there is a mathematical figure of merit given that seems to correlate pretty well with subjective perception.
Go look!
mashaffer said:Of course I may be full of hooey and maybe there is no mathimatical representation that correlates to ugly sound at all.
http://www.gedlee.com/distortion_perception.htm
!!!!!!!!!!
Look at the second presentation on this page!
I've posted this in several of the distortion threads, but no one is commenting on it. Maybe no one is looking at it?
It may not be *the* answer, but there is a mathematical figure of merit given that seems to correlate pretty well with subjective perception.
Go look!
mashaffer said:
Try this:
http://members.iinet.net.au/~jmai/distortion_perception_files/frame.html
I saved the presentation as html and put in my web-space. Maybe you can see it now, though you may need to use IE.
Thanks Jeff. That worked.
That is very interesting stuff. I confess that I am not familiar with th V series stuff but the overall idea is interesting. They are of course still working with frequency domain analysis which is different than what I am proposing. The correlation results are impressive though. It would be interesting to see what the distortions looked like in the time domain.
That is very interesting stuff. I confess that I am not familiar with th V series stuff but the overall idea is interesting. They are of course still working with frequency domain analysis which is different than what I am proposing. The correlation results are impressive though. It would be interesting to see what the distortions looked like in the time domain.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.