Graphic EQ

Dean
Posts: 89
Joined: Fri Feb 10, 2012 6:11 pm

Re: Graphic EQ

Post by Dean »

bigyo wrote:http://bigyo.untergrund.net/buzz/bigyo_ ... in_v02.zip

compiled using FFTReal library by Laurent de Soras

with SubTickTiming disabled uses much less cpu power, it could run that fast when SubTickTiming is on, but with additional delay...
(edit: I mean Param EQs in this post, not graphic EQs, read the corrected, so either replace the words here or read the corrected post a couple of posts down)

interesting work there sir, is there some reason why the sliders have different divisions comparing your recent Graphic EQ and the even more
recent Linear Graphic EQ? For example, the GEQ has around 5 settings per 1dB gain, for LGEQ it's a solid 2 (1 slider move = 0.5 db)
LGEQ has around 0.07-0.09 Hz per slider movement, for the GEQ, you don't get any .0x digit (only 0.x), but I'd estimate each slide movement
to represent something between 0.02-0.04 Hz. It doesn't really make any difference as far as I'm concerned, but for example, this difference
between the two makes it impossible to copy data from one of the EQs to the other.
Last edited by Dean on Wed Feb 22, 2012 3:23 pm, edited 1 time in total.
mute
Posts: 417
Joined: Mon Nov 21, 2011 8:30 pm

Re: Graphic EQ

Post by mute »

The way they work is different. Graphic equalizers for simplicity use static shelving filters based on the low/mid/high bands you are boosting or cutting.

This linear phase eq uses some sort of secondorder filter where you set the (center?) frequency and adjust the Q (width) for the amount of frequency response per band. In practice, this is more similar to a parametric EQ than the Graphic EQ's that started this thread. I won't be shocked if in a future version, we'll be able to select the filter type for each band (low/mid/high bandpass?)...

Speaking of the filter setup.. mind elaborating a bit on what each band is doing on eqlin, bigyo?
Dean
Posts: 89
Joined: Fri Feb 10, 2012 6:11 pm

Re: Graphic EQ

Post by Dean »

mute wrote:The way they work is different. Graphic equalizers for simplicity use static shelving filters based on the low/mid/high bands you are boosting or cutting.

This linear phase eq uses some sort of secondorder filter where you set the (center?) frequency and adjust the Q (width) for the amount of frequency response per band. In practice, this is more similar to a parametric EQ than the Graphic EQ's that started this thread. I won't be shocked if in a future version, we'll be able to select the filter type for each band (low/mid/high bandpass?)...

Speaking of the filter setup.. mind elaborating a bit on what each band is doing on eqlin, bigyo?
I'm sorry, it was a strange typo on my case (as I never use graphic eqs myself...) I was comparing the Linear ParamEQ to the ParamEQ which was also mentioned in this thread (as being posted to Buzzmachines.com). so what I was trying to say was :

interesting work there sir, is there some reason why the sliders have different divisions comparing your recent ParamEQ and the even more
recent LinearParamEQ? For example, the PEQ has around 5 settings per 1dB gain, for LPEQ it's a solid 2 (1 slider move = 0.5 db)
LPEQ has around 0.07-0.09 Hz per slider movement, for the PEQ, you don't get any .0x digit (only 0.x), but I'd estimate each slide movement
to represent something between 0.02-0.04 Hz. It doesn't really make any difference as far as I'm concerned, but for example, this difference
between the two makes it impossible to copy data from one of the EQs to the other. (of course, as mute mention you can't move shelf filter data as of now,
so I'm just refering to the PeakEQ data)
User avatar
bigyo
Posts: 53
Joined: Tue Jan 17, 2012 2:29 pm
Contact:

Re: Graphic EQ

Post by bigyo »

http://bigyo.untergrund.net/buzz/bigyo_ ... in_v03.zip

Algorithm of this linear phase eq is rather simple. It's a second order iir ParamEQ converted to zero-phase fir-filter using forward-backward filtering. So ParamEQLin should sound similar to ParamEQ, except phase response of course. There is an attribute to set filter length. If the filter is not long enough, it's not accurate for low frequencies, but longer filter means more latency.

Sliders are now compatible with ParamEQ. Cpu usage is reasonable, but there is an extra latency.
I think i'll add this as an option, that is less cpu with more latency or more cpu with less latency.
Dean
Posts: 89
Joined: Fri Feb 10, 2012 6:11 pm

Re: Graphic EQ

Post by Dean »

bigyo wrote:http://bigyo.untergrund.net/buzz/bigyo_ ... in_v03.zip

Algorithm of this linear phase eq is rather simple. It's a second order iir ParamEQ converted to zero-phase fir-filter using forward-backward filtering. So ParamEQLin should sound similar to ParamEQ, except phase response of course. There is an attribute to set filter length. If the filter is not long enough, it's not accurate for low frequencies, but longer filter means more latency.

Sliders are now compatible with ParamEQ. Cpu usage is reasonable, but there is an extra latency.
I think i'll add this as an option, that is less cpu with more latency or more cpu with less latency.
sounds great sir :)
User avatar
mantratronic
Posts: 296
Joined: Mon Nov 21, 2011 7:23 pm

Re: Graphic EQ

Post by mantratronic »

have you looked at the recent updates to machineinterface.h that handle delay reporting for the machine? it might be a way round the latency concerns
User avatar
bigyo
Posts: 53
Joined: Tue Jan 17, 2012 2:29 pm
Contact:

Re: Graphic EQ

Post by bigyo »

mantratronic wrote:have you looked at the recent updates to machineinterface.h that handle delay reporting for the machine? it might be a way round the latency concerns
yes, this machine has GetLatency implemented, but more latency means other machines would be delayed more...
User avatar
mantratronic
Posts: 296
Joined: Mon Nov 21, 2011 7:23 pm

Re: Graphic EQ

Post by mantratronic »

Ah, well having used jupiter 3.1, with up to 1.5 seconds of latency, i would be happy with your high latency option then. :oops:

Buzz just pauses for the time needed on start, and even with 1.5 seconds its barely noticable after a few start/stops. Very smooth.

I'm looking forward to hearing it! :)
User avatar
bigyo
Posts: 53
Joined: Tue Jan 17, 2012 2:29 pm
Contact:

Re: Graphic EQ

Post by bigyo »

User avatar
bigyo
Posts: 53
Joined: Tue Jan 17, 2012 2:29 pm
Contact:

Re: Graphic EQ

Post by bigyo »

added inertia and changed the way input signal and impulse response are divided into blocks, so now latency and cpu usage are low.
inertia still uses a lot of cpu. and i think it will be very hard to reduce it.

http://bigyo.untergrund.net/buzz/bigyo_ ... in_v06.zip
User avatar
Evelon
Posts: 147
Joined: Tue Nov 22, 2011 11:00 pm

Re: Graphic EQ

Post by Evelon »

Much appreciated!
User avatar
bigyo
Posts: 53
Joined: Tue Jan 17, 2012 2:29 pm
Contact:

Re: Graphic EQ

Post by bigyo »

User avatar
Klangkulisse
Posts: 304
Joined: Tue Nov 22, 2011 12:20 am
Location: ••• Düsseldorf ••• Made of Light

Re: Graphic EQ

Post by Klangkulisse »

Thanks a lot. I like the flexible track adding very much. Cool.
User avatar
Joachip
Posts: 352
Joined: Mon Nov 21, 2011 7:21 pm
Contact:

Re: Graphic EQ

Post by Joachip »

ParamEQlin is an excellent machine! It behaved very well in my first little test.
I think it would be handy if it defaulted to 2 or 3 bands by default, but that's a minor thing.
Post Reply