Hey all,
Just wanted to invite those in the Csound/Cabbage community to check out this spectral delay instrument I created - inspired by the Native Instruments classic. It’s in pretty rough UI form right now - but perfectly functional. Thoughts and suggestions welcome!
Quick video walkthrough:
This sounds beautiful Micah! Thanks for sharing.
Really great - exactly the sort of sound processing that I got into Csound for.
This is great. All of your Csound-based, Cabbage-built, plugins are AWESOME. This one is even more special. Thanks so much for all the inspiration.
Thanks Dr B!
Very cool plug-in. Is it basically X bandpass filters, each with a delay assigned to it?
(I know it is an old thread!)
I’d be curious to know as well, but I suspect it uses the pvs opcodes, not bp filters.
[https://csound.com/docs/manual/SpectralRealTime.html]
[https://flossmanual.csound.com/sound-modification/fourier-analysis-spectral-processing]
Here’s a really simple example. I imagine if this type of idea was set up with arrays/tables & loops it could work:
<CsoundSynthesizer>
<CsOptions>
-odac
; -o 0_spectral_delay.wav
</CsOptions>
<CsInstruments>
sr = 48000
ksmps = 32
nchnls = 2
0dbfs = 1
iSaw = ftgen(0, 0, 1024, -7, 1, 1024, -1)
giSaw = ftgen(0, 0, 1024, 30, iSaw, 1, 30)
instr trig
iArr[] fillarray 45, 47, 49, 52, 56, 57, 59
kTrig = metro(1)
kNote = iArr[int(trandom:k(kTrig, 0, 6.999))]
schedkwhen(kTrig, 0, 0, 1, 0, 5, kNote)
endin
instr 1
aEnv = expon:a(.7, .4, .001)
aSaw = poscil(aEnv, mtof:i(p4), giSaw)
fSig = pvsanal(aSaw, 128, 128/4, 128, 1)
kBin1 = 7
kBin2 = 4
kAmp1, kFrq1 pvsbin fSig, kBin1
kAmp2, kFrq2 pvsbin fSig, kBin2
aL = poscil3(kAmp1 * 4, kFrq1)
aR = poscil3(kAmp2 * 4, kFrq2)
aDelL init 0
aDelR init 0
iFdbk = .7
aDelL = delay(aL + aDelL * iFdbk, .360)
aDelR = delay(aR + aDelR * iFdbk, .490)
outs aSaw + aDelL, aSaw + aDelR
endin
</CsInstruments>
<CsScore>
i"trig" 0 16
</CsScore>
</CsoundSynthesizer>
The one issue here is reading the bins directly creates alot of noise from adjoining bins. If you instead use pvsarp you can independently amplify a bin while attenuating the surrounding ones.
instr 1
aEnv = expon:a(.7, .4, .001)
aSig = poscil(aEnv, mtof:i(p4), giSaw)
; pvsanal ain, ifftsize, ioverlap, iwinsize, iwinshape
fSig = pvsanal(aSig, 128, 128/4, 128, 1)
kBin1 = 1/64 * 7
kBin2 = 1/64 * 4
fSig1 pvsarp fSig, kBin1, 1, 5
fSig2 pvsarp fSig, kBin2, 1, 5
aL pvsynth fSig1
aR pvsynth fSig2
aDelL init 0
aDelR init 0
iFdbk = .7
aDelL = delay(aL + aDelL * iFdbk, .360)
aDelR = delay(aR + aDelR * iFdbk, .490)
outs aSig + aDelL, aSig + aDelR
endin
You can hear & see a pretty audible difference.
Likely there are a few different ways one might implement this idea, probably more efficient than this.
I think this the GitHub for the project: https://github.com/micah-frank-studio/Lore/blob/main/GenerativeDevices/SpectralDelay/spectralDelay.csd
Here’s the GitHub for it
Not sure if it works with the lastest cabbage though so it might need some modification.
I’m curious, what are you using to visualize the audio? Thanks 
It’s just Cabbage’s built-in Spectrogram
https://cabbageaudio.com/docs/signaldisplay/
I ended support and development for it just because it was too bloated. In a was kind of an experiment for me to test the limits of Csound and Cabbage but the customer support for it became too much. The product page still lives here: https://puremagnetik.com/products/lore-experimental-sound-console
I do plan to break the spectral part out into its own device though as it works quite well
We’re going a little off-topic here, but I can’t believe how much mileage @ST_Music gets from Csound on Android. Are you still using only Android for all your Csound needs?
Actually I just got an older Macbook so really looking forward to diving into using Cabbage on that. Never had a Mac before, only PC, so still figuring out how to navigate the OS. But Cabbage installed and seems to be running ok
If you got this far on your Csound journey with a phone, I’m really excited to see what you can up with a Macbook!