KVR Audio

Cruba · Post by **Cruba** » Fri Nov 17, 2017 1:24 pm

Hi all!

I came up with the idea, if there is a smaller unit than Herz to make minimal frequency slices.

Let's say you have Bassdrum & Bass from 20 - 200 Hz.

Full frequency spectrum would be xoxoxoxoxo

The Bassdrum get x_x_x_x_x_
The Bass get _o_o_o_o_o

Both instruments would have half of the full frequency spectrum, but in microslices and so I think would not notice, something is missing.

Dunno if it is just a dumb idea or something that's still aviable.

resynthesis · Post by **resynthesis** » Fri Nov 17, 2017 1:34 pm

You'd probably need incredibly steep filters with their concomitant artifacts.

Cruba · Post by **Cruba** » Fri Nov 17, 2017 1:41 pm

Well, what I need is eq that has let's say 10000 steps between 20 Hz and 200 Hz and then skip each 1 3 5 7 9 and so on step and the other skip 2 4 6 8 10...

We have Cockos ReaEQ, which can do unlimited bands, but it would be hard work do program it and I also don't know if the architecture kills cpu at last.

resynthesis · Post by **resynthesis** » Fri Nov 17, 2017 1:51 pm

Yes but that means cutting out 0.018Hz slices between 20 and 200Hz if you could use an infinite slope. Maybe some sort of resynthesis would be more reasonable

DaveClark · Post by **DaveClark** » Fri Nov 17, 2017 6:07 pm

Hi Cruba,

I'm not aware of any such frequency vst. However, I did write a program that used a similar idea to separate instruments in a centered recording. The entire recording was FFT'd using a very large window (offline processing). For each bin, if the magnitude of (say) the first instrument was larger than the second, that instrument was assigned to the left channel; if the magnitude of the second instrument was larger than the first, it was assigned to the right channel. This did a far better job of separating signals than expected.

Based on these results, I would advise anyone writing a routine such as you suggested to use the relative magnitudes rather than alternating blindly. In other words, for each bin, if the "bassdrum" magnitude is larger, assign to a "bassdrum" array and do "bassdrum" processing; if the "bass" magnitude is larger, assign to a "bass" array and do "bass" processing.

Of course, this advice requires the programmer to make some kind of determination as to which signal is which, and this may not be easy. But alternating the signals such as you described will produce two signals that will essentially be equivalent in information to two down-sampled signals that sound the same. If the underscores are zeros in your description "x_x_x_x_..." then it's more like two nearly identical half-sample-rate signals that are being up-sampled. Needless to say, these signals would be very highly correlated. I don't see much use for such a thing in music, but perhaps there is.

Regards,
Dave Clark

MogwaiBoy · Post by **MogwaiBoy** » Fri Nov 17, 2017 7:34 pm

Before you try 10,000 alternating cuts between 20 and 20,000hz... why don't you try 10 for starters and see how it sounds?

Or you could try 2 instances of Waves GEQ and cut even or odd sliders.

Cruba · Post by **Cruba** » Fri Nov 17, 2017 8:27 pm

Detailed infos Dave, ty!

Why to do it? Simple: I wanna put Bassdrum & Bass together in same frequency range and if you do the XOXOXO game, I hope you have something better than ducking etc.

Mog: I try a test with Cockos Rea EQ, hope it's possible to set 500 Bands from 60 - 160 Hz. Hope CPU and Rea like 500 Bands...

funky lime · Post by **funky lime** » Fri Nov 17, 2017 8:56 pm

Cruba wrote:Detailed infos Dave, ty!

Why to do it? Simple: I wanna put Bassdrum & Bass together in same frequency range and if you do the XOXOXO game, I hope you have something better than ducking etc.

Mog: I try a test with Cockos Rea EQ, hope it's possible to set 500 Bands from 60 - 160 Hz. Hope CPU and Rea like 500 Bands...

The filters aren't steep enough, and I'd guess all you'll end up doing is screwing up the phase.

Maybe try EngineersFilter (freebie) by RS-MET?

Chandlerhimself · Post by **Chandlerhimself** » Sat Nov 18, 2017 2:38 am

You can do something similar to this with MSpectraldynamics. It will use FFT to break a signal up into bands and then you can insert a sidechain input. From there you can duck the main signal using the sidechain, but it will only duck the frequencies present in the sidechain. It works well IMO.

Of course this is slightly different from your idea, but honestly I think your idea would sound like comb-filtering when both instruments aren't playing at the same time.

Ichad.c · Post by **Ichad.c** » Sat Nov 18, 2017 5:20 am

Just use 2x flanger/delay lines(comb filter), 50% mix, no modulation, no feedback. And Chandlerhimself is correct, will sound like static comb filtering.

Cruba · Post by **Cruba** » Sat Nov 18, 2017 7:25 am

Wow... Thank you for all the professional input... I was afraid I'll be the idiot asking this, but KVR shows up really friendly and professional users again!

Chandler:

Well, the basic question all in all is: Would the ear recognize a difference if there is a tool that does my idea?

Sugar Audio · Post by **Sugar Audio** » Sat Nov 18, 2017 10:29 am

funky lime wrote:
Cruba wrote:...The filters aren't steep enough, and I'd guess all you'll end up doing is screwing up the phase. ...

Filterizor is able to create up to 100 20th order band / band pass / notch order filters in one instance. Using zero phase mode the result will be phase neutral and cpu friedly. Anyway I'm not sure what this experiment would sound like.

DJ Warmonger · Post by **DJ Warmonger** » Sat Nov 18, 2017 10:47 am

What you describe is similiar to comb filter, but the actual outcome might not be exactly what you expect.

djscorb · Post by **djscorb** » Sat Nov 18, 2017 12:01 pm

Better to allocate sounds that share frequency ranges to different points in time. Then you retain purity and power, dynamics and contrast.

I can't remember where I heard it, but the phrase; "When the kick and bass fight, the kick must win the battle" is a great mantra when tackling this issue.

There are a million and one tricks to get kick and bass working in synergy. I wouldn't want either of these all important mix elements to be 50% leaner, just so that when they occur together (probably a quarter of the time at most), they interlock with less peaks or cancellation. Seems like too much for of a trade off to me.

Interesting thinking though!

Cheers

Scorb

Chandlerhimself · Post by **Chandlerhimself** » Sun Nov 19, 2017 4:27 am

Cruba wrote:Wow... Thank you for all the professional input... I was afraid I'll be the idiot asking this, but KVR shows up really friendly and professional users again!

Chandler:

Well, the basic question all in all is: Would the ear recognize a difference if there is a tool that does my idea?

Yes, it would either significantly thin out the sound or have no effect at all. A sound like bass only occupies a few frequencies, which sound match the harmonic series(excluding the transients). Either your eq would miss the frequency of the bass and have no effect or remove the bass sound leaving it thin.

Is such a frequency vst aviable?