Author Topic: New filters  (Read 84067 times)

Xavier

  • Administrator
  • Hero Member
  • *
  • Posts: 2257
    • View Profile
Re: New filters
« Reply #120 on: December 15, 2019, 06:28:00 PM »
I went through all the filters.
Thanks again for al the work you put into this.  :)

Here are the notes i wrote down, please let me know if they're expected behaviour or algo problems.

BP2 : Q? Just increase volume ?

Peak : Res : usefull between .8 and 1.0
 Res: something strange when changed from .99 to 1.00

Notch : Freq Max = 8Khz ?
Res : Small impact

BpDs : instable at Res = 1.0

Pann : pos = panning ? not sure. Lots of frequency impact from 0 to 1!
Spread = ?

Sat : digital noise. Is it how it's suposed to sound ?

LpSn : What does Pos do ?
Inverse both to target Freq in the matrix ?

Note4 : low pass filter on Freq param ?

18db : res smaller when freq smaller

La+d I like this one, are the other ones usefull ? This one seems to sound better.


Also this recent commit worries me :
https://github.com/pvig/preenfm2/commit/158e858c436270d610b4066bbafeebea9fc6180b
Removing velocity which was a value between 0 and 127 since the beggning can only make a big changes to all those filters.
As everybody tried and liked the filters before this change, i'm wondering the impact this has.



« Last Edit: December 15, 2019, 07:02:31 PM by Xavier »

Toltekradiation

  • Sr. Member
  • ****
  • Posts: 254
    • View Profile
Re: New filters
« Reply #121 on: December 15, 2019, 09:06:05 PM »
hi Xavier,
don't worry about the recent commit : i tried to make allpass filter frequency velocity dependant, was not useful, it was a regression ; last commit just put it back to a previous state.


BP2 : yes it doesn't work very well, it was a attempt to double sampling the original algo.

Peak: ok, i modified the reso curve response. there a small click when reso = 1, could not get rid of yet...

Notch: the resonance has expected behaviour, but, yes , this algo don't go very very high

BpDs : self oscillation is not expected, i modified this in my last commit

Pann : this one is a mixture of panning and 3x lowpass with different frequency spread around f

Sat : yes, it is a very simple digital distortion, sounds as expected.

LpSn : pos is an offset for a transfer function, add distortion, depending on pos and number of voices played. i let param1 control it to allow variable distortion. try it on a pure sin poly patch to use it a its best.

Not4 : quad notch filters, use it with a slow lfo for phasing effect

18db : ok, fixed this in last commit

La+d: ok, yes the ladder filter department is quite crowded, why not removing Lad+, Diod and L+d+

about the h3o+ (tb like) : i reworked the dynamics and accent circuit in one of my last commit if you want to check it out.

Xavier

  • Administrator
  • Hero Member
  • *
  • Posts: 2257
    • View Profile
Re: New filters
« Reply #122 on: December 15, 2019, 10:19:48 PM »
Cool, thanks for your reply.

I think i won't integrate the following :
. Xor, Txr1, Txr2, LPx1, LPx2 : Cannot get anything else than digital noise.
. BP2 : "does not work very well". Not sure it brings something interesting beside the other BP.
. Lad+, Diod and L+d+ : I maybe wrong but i think LA+d is enough and sounded better in my test.
. I hesitate for SAT.

If you or someone else has a strong opinion that i'm making a mistake not taking some of them, please argue and teach me how to get them sounding !
I'll release a new firmware hopefully this year.

I'm also working on the editor.
The new version of the editor will have the filter list in a xml file so people will be able to use your firmware with as many filters as you want.
Only the xml will have to be modified to have a working editor with additional filters.
You can continue your experiences  :)
« Last Edit: December 15, 2019, 10:23:11 PM by Xavier »

observer

  • Team member
  • *
  • Posts: 37
    • View Profile
Re: New filters
« Reply #123 on: December 17, 2019, 11:59:31 AM »
I agree. Some of these filters are a lot more useful than others. I'd rather not have to scroll through a long list to get to the ones that give me anything other than glitched signals. Quality over quantity any day.

maki

  • Team member
  • *
  • Posts: 26
    • View Profile
Re: New filters
« Reply #124 on: December 24, 2019, 03:27:51 PM »
Toltekradiation, thank you for everything! If you are done with filters and still have some extra energy to work, could you then make:
1)a tracking generator for each slot in Modulation Matrix?
2)paraphonic properties in Preenfm2? A counter counts the number of held/active notes and this value changes something in a modmatrix output.
3)Or you could subtract notenumber of those held notes/the time difference between the held notes and use those values further in modmatrix outputs.
4)Or you could use the speed of incomming note on/cc events to change something in modmatrix output.
Maki

Toltekradiation

  • Sr. Member
  • ****
  • Posts: 254
    • View Profile
Re: New filters
« Reply #125 on: December 28, 2019, 11:27:28 PM »
hello @maki,
unfortunatelly, i have not much energy to continue, my day job take most of my time for now, maybe later.
Anyway you have very nice ideas to be tested ; very expressive possibilities  :)

baxgas

  • Team member
  • *
  • Posts: 22
    • View Profile
Re: New filters
« Reply #126 on: January 06, 2020, 10:20:14 PM »

. I hesitate for SAT.


@Xavier – I find Sat very nice on pure Sine-waves, can make nice Waveshaping.

Didn't have time recently to try out the other algos, but you are right about Txr1, Txr2 didn't find those very useful.
The different pole HP, LP, BP filters have subtle differences, that's true, maybe only the max. Res setting
LPx variations are also a bit so-so, but they can be useful to have multiple filters if possible.

Altogether it is definitely good to have the filters implemented in the official firmware, Toltekradiation did a wonderful work!

cube48

  • Team member
  • *
  • Posts: 14
    • View Profile
Re: New filters
« Reply #127 on: January 14, 2020, 08:52:02 PM »
Thanks to everyone who contributed into this thread and especially to Toltekradiation and Xavier for bringing in new filters. I just learned recently about this development and it made me to get my 3rd PreenFM2 (sold the first two but always come back to this eternal love affair  ;D).
They sound really good and profoundly enhance the sonic potential.

After checking them all I've realized it would be great to have the second filter parameter available as destination in mod matrix as well because on some filter types it seems the cutoff sits on the second position (correct me if I'm wrong). But having the resonance (or others) under modulation control would be awesome. I have a vague feeling there was some performance limitation in the past that didn't allow this. Has anything changed with new, more efficient filters in this regard?

Edit: I forgot to add that I love now how the Gain parameter distorts the signal nicely when cranked up. Great work!

Edit2: OK, found the answer here.
« Last Edit: January 15, 2020, 12:48:53 AM by cube48 »

solipsvs

  • Team Member
  • ***
  • Posts: 222
    • View Profile
Re: New filters
« Reply #128 on: May 06, 2020, 09:14:12 PM »
is there a possibility to add a comb filter?

Toltekradiation

  • Sr. Member
  • ****
  • Posts: 254
    • View Profile
Re: New filters
« Reply #129 on: May 07, 2020, 10:17:05 AM »
hi,
i don't think so, because this kind of filters need quite some memory to work ; it use delays.
I will check what is possible to do.