Kevinofnine Debut Ensembles

Just a little note to tell you all about our new website and debut Reaktor Ensemble.

It's called the KEVINOFNINE EQ BUNDLE. We're new around here so we thought we'd kick it off with a good start.

You can download the bundle and find out more about it here:

http://kevinofnine.on.to/

All the best,

The KEVINOFNINE team.

 

One post is enough.

 

Also.. why winrar? Why not stick it in the user library?

It's obvious you put a lot of work in it... put it in a place where more people will get a chance to enjoy it.

Also, could I suggest implementing your EQ via a drawable EQ curve? Adjusting 108 faders and 108 knobs would seriously aggrivate my psuedo-RSI.

 

Thanks for the ideas.

Thanks John. The drawable curve is a good idea. We did not even think of that. The thing is being able to fit it all on one screen was perhaps too important upon reflection. Also, the trade off for having so many knobs and faders is having the flexibility instantly available, while actually using all 108 knobs and faders would be redundant in a sense, since the Q's are variable.

Winrar seems to shrink the files down incredibly, but I'm sure Winzip does a good job as well. We were going to upload it to the library if there was any initial interest in it.

Thanks again for the reply. We posted it twice as I thought I had posted it in the wrong forum initially. I thought it was supposed to be in the 'user library' section.

Cheers.

 

Ideally if you have that many bands, the Q for each should be fixed in a manner so you could just draw in the curve and have everything represented properly. If you drew a curve and had user set Q for each, then the actual curve would look much different.

However each frequency band is a different size for what you're doing, hence you have to figure out the Q for -each band- in order to do this properly.

Basically... IMO, such sorts of EQing is best left to FFT algorithms.

 

I see what you are saying.

Yes, I see what you are saying with the Q. I thought about scaling the Q with the fixed bandwidths for the 9 octaves. There is one downside with that approach. Not being able to modify the Q for each band or limiting it to its respective/applicable 'range' leaves you with less control overall. Being able to span further from the particular centre frequency and affecting other frequencies in turn yields different and more flexible results, as opposed to cutting or boosting from additional neighbouring bands to achieve "the same" result.

For example: 2 bands an octave apart, boosted that have bandwidths that overlap anywhere in between the 2 bands, will be different than the same 2 bands boosted with fixed Qs that span their respective semitone frequency ranges, plus other bands in between to 'emulate' the same boosted overlap that occurs with the bands with variable Qs. With the first approach you can also set the Q's for each band to equal about a semitone, so this option was left available for people.


Though the differences in the results may be subtle in the 2 approaches I find that it makes a big different on instruments such as acoustic guitars, or in particular stringed instruments that carry complex overtones/harmonics. Drawing the curve out with each band having its applicable Q that is fixed is great idea though. Having control over each Q must be more flexible in the end, but the fixed Q idea would be nonetheless extremely useful and accurate as well. The EQ is not really supposed to be a quick and dirty process, but I can understand what you mean about the amount of controls. The main idea was to create a Peak EQ that saves time skipping mundane calculations between pitch and frequency values, and to relate the simple power of an EQ directly to the music that it is processing.

For us it's a dream, but I can see why it can be a 'nightmare'.

Thanks again for the input. I think I will get started on your idea for this soon. It would be great to include it in the ensemble or its own ensemble as an option.

All the best.

 

The amount of filters it takes to do that for that many bands makes it impossible to run on my system. I was doing something similar... actually using spectral analysis via a series of filters to do some vocoding. My computer crapped out on me though.

Instead of using the peak EQ module, you might get better results using 1 lowpass and 1 highpass to cut out each band. Keep in mind how they overlap though. If you boost one, then boost the one next do it, depending how how big the overlap is, the overlaped area could be made louder than it should be, or you'll have a notch.

Doing this properly really is a nightmare. There's a reason you don't see any systems with drawable EQ curves done entirely with a series of filters. It's impossible to implement transparently, and also eats a huge about of CPU power.