Is this MSP patch "amplitude gate" possible in PD extended?

Reply to Is this MSP patch "amplitude gate" possible in PD extended? on Thu, 15 Oct 2015 20:50:00 GMT

gummi — Thu, 15 Oct 2015 20:50:00 GMT

I recently watched a youtube vid of someone using fast fourier transforms to manipulate audio in a very specific way; he called it an "amplitude gate" :

I am not necessarily wanting someone to do the work for me in translating this MSP patch to PD, but I really want to utilize this patch in my work. I am curious - for anyone with knowledge of both MSP and PD - if the libraries and extensions in PD cover the needs that this patch has, and what they would entail. Hopefully that isn't a demanding task, as it seems quite simple.

Also I am curious if there is a more commonplace term than "amplitude gate", and if this patch is already abstracted in a PD depository under a different name. I think this patch has quite a lot of possibility!

Reply to Is this MSP patch "amplitude gate" possible in PD extended? on Thu, 15 Oct 2015 21:25:13 GMT

gummi — Thu, 15 Oct 2015 21:25:13 GMT

Mainly curious due to PFFT. I haven't explored much with FFT work in pd yet, but I've heard that this is one area where MSP shines in terms of ease and use and efficient computation.

Reply to Is this MSP patch "amplitude gate" possible in PD extended? on Thu, 15 Oct 2015 21:48:38 GMT

baptiste_c — Thu, 15 Oct 2015 21:48:38 GMT

http://www.pd-tutorial.com/english/ch03s08.html
You'll find an example of spectral delay. It should be possible to turn it into a spectral gate.

Reply to Is this MSP patch "amplitude gate" possible in PD extended? on Sat, 17 Oct 2015 08:37:32 GMT

gsagostinho — Sat, 17 Oct 2015 08:37:32 GMT

Hi there,

As you saw in that video, the object [pfft~] is mainly a container (although it automatically takes care of block size, window type, etc. for you) and the magic actually happens when you use [fftin~] and [fftout~]. In Pd, we don't have a container object like that (although you can use a subpatch or abstraction of course) and the equivalents of [fftin~] and [fftout~] are [rfft~] and [rifft~].

I recreated that patch of the video here, have a look:

example of spectral gate.pd

Take care,
Gilberto

Reply to Is this MSP patch "amplitude gate" possible in PD extended? on Fri, 16 Oct 2015 15:25:49 GMT

gsagostinho — Fri, 16 Oct 2015 15:25:49 GMT

By the way, you may have to play around with the scale of those horizontal sliders in the main patch.

Reply to Is this MSP patch "amplitude gate" possible in PD extended? on Fri, 16 Oct 2015 21:05:43 GMT

gsagostinho — Fri, 16 Oct 2015 21:05:43 GMT

Small update: it turns out [cartopol~] has a bug in which the rightmost outlet outputs a wrong value (-1 times the correct value). So if using pd-extended one must compensate for that (the bug isn't present in Pd-l2Ork, whic his why I didn't find it out immediately).

Anyway, I created an abstraction called [specgate~] for doing exactly what we have been talking here. It is available for both pd-extended and l2Ork in the following link: https://github.com/gilbertohasnofb/pd-abstractions-and-libraries

Cheers,
Gilberto

Reply to Is this MSP patch "amplitude gate" possible in PD extended? on Fri, 16 Oct 2015 21:35:01 GMT

seb-harmonik.ar — Fri, 16 Oct 2015 21:35:01 GMT

@gsagostinho it might be prudent to window the fft as well.

Reply to Is this MSP patch "amplitude gate" possible in PD extended? on Fri, 16 Oct 2015 22:45:19 GMT

gsagostinho — Fri, 16 Oct 2015 22:45:19 GMT

@seb-harmonik-ar The problem I had when using a Hanning window is that if I would set the arguments of the abstraction to 0 and 10000 -- that is, passing only frequencies with amplitudes larger than 0 and smaller than 10000, which is effectively allowing all signals through -- the original signal wasn't being reconstructed properly, it was full of artifacts. This effect could also be perceived if the values are closer to the extremes mentioned above (that is, if the sound is close to its original form).

Reply to Is this MSP patch "amplitude gate" possible in PD extended? on Sat, 17 Oct 2015 09:23:09 GMT

gsagostinho — Sat, 17 Oct 2015 09:23:09 GMT

Please ignore my last comment, the window function wasn't being created properly due to my own noobness, which is the reason the signals were being affected. I have updated the abstractions in my repository and they now have a proper window function:

Reply to Is this MSP patch "amplitude gate" possible in PD extended? on Fri, 23 Oct 2015 07:07:50 GMT

gummi — Fri, 23 Oct 2015 07:07:50 GMT

I keep getting a lot of

"______: no such object"

errors when trying to open the specgate~.pd file, do I need anything special to open it? The original specgate file works, but as you said it has a missing piece and outputs sound incorrectly

Reply to Is this MSP patch "amplitude gate" possible in PD extended? on Fri, 23 Oct 2015 12:42:00 GMT

gsagostinho — Fri, 23 Oct 2015 12:42:00 GMT

I keep getting a lot of "______: no such object" errors when trying to open the specgate~.pd file, do I need anything special to open it?

I don't know, I just tested again my [specgate~] with pd-extended and it works flawlessly here. Which OS and which pd version are you using?

The original specgate file works, but as you said it has a missing piece and outputs sound incorrectly

Well, you can do a little hack to fix it (just like I did with the abstraction [specgate~] for pd-extended):

Just multiply the rightmost inlet of [cartopol~] by -1 and you should get rid of the sound problems.

But please note this is a hack, if you are serious about doing FFT in pd-extended I would recommend you to update your cyclone library to version the most recent version (alpha57 ): https://puredata.info/downloads/cyclone/releases/0.1-alpha57

Cheers,
Gilberto