KVR Audio

93143 · Post by **93143** » Thu Sep 27, 2007 1:32 am

William Sharkey wrote:
rbet wrote:

That is possible. The legs of the "table" just represent the fact that those frequencies are being folded (or shifted) and that the db is being reduced there. The leg on the right could be done by aliasing. And the leg on the left would need to be processed by some fft-ifft resynth thing.

Well, no, it's not, because that shape isn't a function (the value of the dependent variable isn't unique for a given value of the independent variable). Bode plots just don't work that way.

Maybe I'm missing a joke...?

Zaphod (giancarlo) · Post by **Zaphod (giancarlo)** » Thu Sep 27, 2007 4:16 am

93143 wrote:
Maybe I'm missing a joke...?

a fantascientific joke..

MaliceX · Post by **MaliceX** » Thu Sep 27, 2007 4:31 am

Now that I look at that graph once again..it had me thinking.

If that 'were' possible, it'd be more or less an EQ with the ability to remove certain frequencies at exact volume levels.

Too highly complicated, but it doesn't sound impossible. (ie: Keep the loud bass, while you get rid of the quiet ones.)

Zaphod (giancarlo) · Post by **Zaphod (giancarlo)** » Thu Sep 27, 2007 4:34 am

yes, a common tecnique used in noise reduction tools.... even in dataloss encryptions...

MaliceX · Post by **MaliceX** » Thu Sep 27, 2007 4:50 am

Yes but its use in 'lossy' audio compression formats isn't exactly 'user-controllable' to the point you can use a graph and 'tell' it what frequencies you want removed. Plus, their use in those cases is just to 'downsize' files as much as possible without compromising too much quality.

If it were in the case of an EQ, this would be something else, though again what would the practicalities of this be? An extra layer to remove unwanted sounds without interrupting existing frequencies? I don't know, it's just speculation. But yeah, selective hearing.

HunterKiller · Post by **HunterKiller** » Thu Sep 27, 2007 5:16 am

Conclusion, the most commonly used smiley in this thread is Mr

Zaphod (giancarlo) · Post by **Zaphod (giancarlo)** » Thu Sep 27, 2007 7:08 am

HunterKiller wrote:Conclusion, the most commonly used smiley in this thread is Mr

93143 · Post by **93143** » Thu Sep 27, 2007 10:19 pm

Everyone relax - the picture was and remains just a good joke, and it really is a simple mathematical impossibility.

The selective noise-reduction you've been discussing sounds at first like it might make sense of the graph, but when you think about it, you realize that the representation of frequency response as a Bode plot doesn't admit that interpretation. It has to represent a linear transfer function; there's no way to justify a nonfunctional curve.

If you were to reinterpret the vertical scale and re-represent the frequency response as a 2D surface function depending on frequency AND amplitude, then you could represent a denoiser. Oversimplified, it could even look similar to what you've got there.

Add another dimension and you could represent a distortion...

aeron · Post by **aeron** » Fri Sep 28, 2007 3:01 am

Haha, oh wow kvr..

Zaphod (giancarlo) · Post by **Zaphod (giancarlo)** » Fri Sep 28, 2007 4:42 am

aeron wrote:Haha, oh wow kvr..

I'm falling in love for this smiley

Cordelia · Post by **Cordelia** » Fri Sep 28, 2007 5:46 am

Hey, while I've got some smart types here-
Hijack, I know, I didn't get an answer to an earlier post.

What does "Q" stand for, exactly?
What is the relationship between "width" (as in fractions of an octave) and Q?
Higher Q = smaller width, I know, but is there a mathmatical relationship?
I'm trying to replicate settings from NoNameEQ (which has "width"), on other EQs (most of which use "Q").
So, for instance, a .50 setting on NoNameEQ (1/2 octave) would equal what Q setting?
Anyone know?
Thanks.

Muzik 4 Machines · Post by **Muzik 4 Machines** » Fri Sep 28, 2007 6:02 am

i would venture a guess, based on nothing
0.5=2

Cordelia · Post by **Cordelia** » Fri Sep 28, 2007 7:04 am

Ha ha. Thanks, Muzik 4 Machines.
Sorry for the temporary derailing of this thread.
Well, OK, it had come down to a discussion of love of emoticons, but still.
Anyway, Dozius answered my question in another thread.
Carry on...

MaliceX · Post by **MaliceX** » Fri Sep 28, 2007 7:09 am

93143 wrote:Add another dimension and you could represent a distortion...

That'd be the correct term it seems, in this case the graph would literally do that at a selected frequency, and at different levels. For that to be done, would involve far too many algorithmic sequences, and in the end wouldn't sound good at all anyway.

Zaphod (giancarlo) · Post by **Zaphod (giancarlo)** » Fri Sep 28, 2007 7:11 am

93143 wrote:
Add another dimension and you could represent a distortion...

I missed it

million dollar scientific EQ ? [Thread from 2007 bumped]