[PD-announce] Re: pulqui bump version

Hi Roman,

> I'm interested to know why this filter is suitable for multi-band 
> dynamic processing?

citing from the comments of the code I implemented (first example in 
https://www.musicdsp.org/en/latest/Filters/266-4th-order-linkwitz-riley-filters.html 
)

"The sum of the Linkwitz-Riley (Butterworth squared) HP and LP outputs, 
will result an all-
pass filter at Fc and flat magnitude response - close to ideal for 
crossovers."

also note here https://en.wikipedia.org/wiki/Audio_crossover#Overview it 
shows that the "LR2 sum" (Linkwitz-Riley (Butterworth squared)) is flat 
compared to Butterworth filters. (see pic)

> I once created a 4-band multi-band compressor for netpd¹ and I used 
> iemlib butterworth filters for the crossover. 

i also tried this approach but felt that something was missing when 
comparing input to summed output.

> Can I just chain many [pqcrossover~] objects for separating more than 
> two bands?

yes. i tested this with nice results (but noted (measured with [VU]) 
that the summed output is a little bit louder than the input.


  |
  |
[cross 1000   ]
  |           |
  |           |
[cross 100]  [cross 7000 ]
  |       |    |         |
  |       |    |         |
low  mid-low  mid-high  high


also i had just tested this and seems fine to my ears:

  |
  |
[cross 500   ]
  |           |
  |           |
  |           [cross 5000 ]
  |            |         |
  |            |         |
low          mid       high


PS: in a fantastic world i would like to implement in Pd the code from 
"Calf X-Over $ Band" from https://calf-studio-gear.org/#tools

:)


-- 
Mensaje telepatico asistido por maquinas

On 28/01/2025 07:19, Roman Haefeli wrote:
> Hey Lucas
>
> Thanks for sharing.
>
> On Tue, 2025-01-28 at 05:29 -0300, Lucas Cordiviola wrote:
>
>
>> version 0.4.0 adds 2 new objects:
>>
>> - [pqcrossover~]
>>
>>     - 4th order Linkwitz-Riley filters to be able to do multi-band
>> limiting (or other processing)
>
> I'm interested to know why this filter is suitable for multi-band dynamic processing? I once created a 4-band multi-band compressor for netpd¹ and I used iemlib butterworth filters for the crossover. I haven't read much about crossover design and just tuned the band bounds by ear, so that the overall frequency response sounded more or less even to my ears (very naive approach, but I liked the result). Once I found a vanilla butterworth implementation, I tried to replace the iemlib filters, but I wasn't able to make it sound "even", so I gave up.
>
> Can I just chain many [pqcrossover~] objects for separating more than two bands?
>
> Thanks, Roman
>
>
> ¹[http://netpd.org/instruments/dynlib/](http://netpd.org/instruments/dynlib/) (r4mbc)
>
>

_______________________________________________
Pd-announce mailing list -- [email protected]
To unsubscribe send an email to [email protected]
---
[email protected] - the Pure Data mailinglist
https://lists.iem.at/hyperkitty/list/[email protected]/message/63ZSPDAVX4SUHAHFW4CW3AMVQTTWDGRW/

To unsubscribe send an email to [email protected] mailing list
UNSUBSCRIBE and account-management -> https://lists.iem.at/