Reply to thread

Message

<blockquote data-quote="Dan Mortensen" data-source="post: 139402" data-attributes="member: 2826">Re: Possible to use RTA to create an EQ curve?Another reason why 1/3 octave is inadequate: If we divide any given 1/3 octave in two so that it is 1/6 octave, and the lower half is +20 and the upper half is -20, the average of them is 0, or unity, which is what the 1/3 octave analyzer would output.A unity filter is not representative of what&#039;s really going on, and if you try to get rid of that nasty +20 peak by lowering your 1/3 octave slider you&#039;ve taken the -20 one further into negative territory.Substitute any random values for those +/- examples, or divide the band further into 1/12 or narrower bandwidths and then mess with the values, and the problem only gets worse in terms of replicating the spectral waveform. And as you look at narrower and narrower bandwidths, is thinking of +/- 20 db adequate or are the variations more wild than that?A whole series of cans of worms open up, if one is trying to exactly replicate a target spectral waveform. And as you imply, the phase/time response does not currently have a way to adjust like frequency does, although they are all tied together and changing one affects the others.That&#039;s not to say that it&#039;s completely impossible do an OK job of replicating something, but it&#039;s tough to do it accurately.However, if someone has time on their hands and is a pretty good audio person who pays attention to more than just EQ response and is a really clever computer programmer, it would seem like a fun and interesting project to try to give Pauly what he wants and would provide an education in the same way that John Meyer learned about building loudspeakers by focusing ever more closely on analysis and performance conditions (for example). I don&#039;t doubt that Don Quixote is working on this now, and that someday someone is going to come really, really close.</blockquote>

[QUOTE="Dan Mortensen, post: 139402, member: 2826"] Re: Possible to use RTA to create an EQ curve? Another reason why 1/3 octave is inadequate: If we divide any given 1/3 octave in two so that it is 1/6 octave, and the lower half is +20 and the upper half is -20, the average of them is 0, or unity, which is what the 1/3 octave analyzer would output. A unity filter is not representative of what's really going on, and if you try to get rid of that nasty +20 peak by lowering your 1/3 octave slider you've taken the -20 one further into negative territory. Substitute any random values for those +/- examples, or divide the band further into 1/12 or narrower bandwidths and then mess with the values, and the problem only gets worse in terms of replicating the spectral waveform. And as you look at narrower and narrower bandwidths, is thinking of +/- 20 db adequate or are the variations more wild than that? A whole series of cans of worms open up, if one is trying to exactly replicate a target spectral waveform. And as you imply, the phase/time response does not currently have a way to adjust like frequency does, although they are all tied together and changing one affects the others. That's not to say that it's completely impossible do an OK job of replicating something, but it's [B]tough[/B] to do it accurately. However, if someone has time on their hands and is a pretty good audio person who pays attention to more than just EQ response and is a really clever computer programmer, it would seem like a fun and interesting project to try to give Pauly what he wants and would provide an education in the same way that John Meyer learned about building loudspeakers by focusing ever more closely on analysis and performance conditions (for example). I don't doubt that Don Quixote is working on this now, and that someday someone is going to come really, really close. [/QUOTE]

Verification