Rossum Panharmonium Resynthesizer

Cwejman, Livewire, TipTop Audio, Doepfer etc... Get your euro on!

Moderators: luketeaford, Joe., lisa, Kent

Post Reply
User avatar
Pailo
Wiggling with Experience
Posts: 326
Joined: Mon Jun 27, 2011 11:10 pm
Location: PNW
Contact:

Post by Pailo » Thu Aug 01, 2019 1:39 pm

Take the STO from Make Noise.
Feed the wave folded wave into the input of Panharmonium.
Take the Sub OUT and patch it into the V/Oct.
I don't know what it is doing but its like a melody appears out of nowhere.

Strange and fun. :hmm:

User avatar
thetwlo
Super Deluxe Wiggler
Posts: 4927
Joined: Wed Apr 16, 2008 4:05 pm
Location: East Bay, CA US

Post by thetwlo » Thu Aug 01, 2019 4:35 pm

here's a quick example, not the best, wasn't going to put it up since it has a lot of reverb. It's just a simple rhythm from the Assimil8or into it:

https://soundcloud.com/user-91119069/panharm-quick-edit

User avatar
JES
Veteran Wiggler
Posts: 648
Joined: Tue Jun 18, 2013 10:03 pm
Location: Montreal

Post by JES » Thu Aug 01, 2019 4:50 pm

The examples coming through are pretty interesting. Please keep posting them. Feel free to use all the reverb you want.

In the meantime, at the risk of triggering some people from earlier in the thread when this was still a "coming soon" sort of deal (<--kidding about the triggering part), I know this isn't a vocoder, but it does seem a lot like some of the software I have that uses phase vocoding. Looking at this explanation: http://www.panix.com/~jens/pvoc-dolson.par it seems to be pretty similar in layout even though there all sorts of additional possibilities (FM, oscillator type not just sine, etc).

Can someone explain to me how it's NOT a relative of a phase vocoder? I would be grateful just for the clarification.

PS -- if you don't know what a phase vocoder is, the above link is a decent explanation. It's not the same thing as a regular vocoder at all, though they're historically related.
"Now, I'm off to hook my cat's litterbox up to an envelope follower." --Aragorn23

User avatar
starthief
Super Deluxe Wiggler
Posts: 4407
Joined: Tue Oct 04, 2016 10:39 am
Location: St Louis, MO
Contact:

Post by starthief » Thu Aug 01, 2019 5:21 pm

I'm not a DSP expert by any means but I would say it's certainly related to a phase vocoder. I believe what it's doing is:

-- using FFT to do the analysis
-- filtering the FFT results with bandpass/notch
-- sorting them by strength and keeping the strongest 33 frequencies
-- feeding those to an oscillator control section with slew on the amplitude (blur) and frequency (glide)
-- oscillators can exponentially be pitch shifted (freq, FM, V/OCT) or detuned relative to each other (warp) and have selectable waveforms
-- internal VCA per oscillator, controlled by spectral data and wave crossfading

Instead of a phase vocoder:

-- use filter banks to do analysis
-- optional: monkey with the levels
-- use oscillator banks to do resynthesis
OR
-- use FFT to do analysis
-- optional: monkey with the data
-- use inverse FFT to do resynthesis

I've used the latter technique in a few VST plugins. Just selectively zeroing amplitude on the weakest or strongest bins can be kind of neat.

User avatar
starthief
Super Deluxe Wiggler
Posts: 4407
Joined: Tue Oct 04, 2016 10:39 am
Location: St Louis, MO
Contact:

Post by starthief » Thu Aug 01, 2019 5:27 pm

Also there may be some extra psychoacoustic wizardry going on in their algorithm -- as in, choosing the most "important" bands rather than necessarily the strongest amplitude ones, according to how the others are clustered.

Or not, I have no idea. :)

User avatar
boramx
lepidopteran orgasm
Posts: 2165
Joined: Sat Mar 20, 2010 12:02 pm
Location: Infant Island

Post by boramx » Fri Aug 02, 2019 7:49 am

i noticed the lag in one of the earlier demos and didn’t seem too bad. hows is it for you current holders?

given what startheif says about the wet stereo - i kinda wonder why it needs to be stereo at all unless the lag is so insignificant that one would use it often as an effect with the dry signal present.

my first thoughts re:stereophony was giving it two rather different signals at LR and getting interesting effects with pure wet results.

User avatar
starthief
Super Deluxe Wiggler
Posts: 4407
Joined: Tue Oct 04, 2016 10:39 am
Location: St Louis, MO
Contact:

Post by starthief » Fri Aug 02, 2019 7:51 am

This is the second time in two days I have tried to post and the post just disappeared into the ether. First time I thought it was my own mistake...

Anyway. A sound example.

http://starthief.net/stuff/panharm1.flac

DPO -> Panharmonium -> delay -> feedback into Panharmonium

Mid/side before and after Panharmonium in the feedback loop.

Just messing with knobs unscripted, with a Marbles sequence. At one point I kind of screw up and freeze Panharmonium but don't silence the DPO which is still sequenced and it's dissonant as hell.

User avatar
starthief
Super Deluxe Wiggler
Posts: 4407
Joined: Tue Oct 04, 2016 10:39 am
Location: St Louis, MO
Contact:

Post by starthief » Fri Aug 02, 2019 8:44 am

boramx wrote:i noticed the lag in one of the earlier demos and didn’t seem too bad. hows is it for you current holders?
Latency seems to be somewhere in the 40ms range, but I think it varies with the slice clock.

http://starthief.net/stuff/panharm_latency.wav

This is a short "tick" from Rings in the left channel, Panharmonium's output in the right channel. No blur or feedback, 33 sine oscillators. I start at the fastest available slice time and turn up the multiplier as it goes along.

(And yeah, it can sound a little weird with inharmonic inputs.)

Given the slice clock's max resolution is 17ms, it almost always sounds better if you slow it down, sync to a musical clock or else blur and feedback it.


Here's one where I'm just triggering Rings strum, syncing Panharmonium to the same clock, and giving Panharmonium Voices and V/OCT some very simple sequences. Panharmonium wet goes through Natural Gate, triggered by its slice clock output. I tweak the feedback, clock multiplier and octave. Not musically sophisticated or anything :hihi:

http://starthief.net/stuff/panharm_simple.flac

User avatar
Mend
Common Wiggler
Posts: 123
Joined: Wed Oct 04, 2017 10:03 am
Location: The Netherlands

Post by Mend » Fri Aug 02, 2019 2:21 pm

starthief wrote:This is the second time in two days I have tried to post and the post just disappeared into the ether. First time I thought it was my own mistake...

Anyway. A sound example.

http://starthief.net/stuff/panharm1.flac
This is beautiful. Going to try something similar.

User avatar
Funky40
Super Deluxe Wiggler
Posts: 5812
Joined: Mon Nov 17, 2008 7:08 pm
Location: on a big voyage

Post by Funky40 » Fri Aug 02, 2019 7:08 pm

starthief wrote: Here's one .........
yeah, feed the people who can´t get one cause of space limitations and/or uncooperative banks.
Its like getting a "dessert surprise" :mrgreen:
For sale:........it won´t sell, so i stopp bother you ;)

User avatar
starthief
Super Deluxe Wiggler
Posts: 4407
Joined: Tue Oct 04, 2016 10:39 am
Location: St Louis, MO
Contact:

Post by starthief » Fri Aug 02, 2019 9:30 pm

A fun patch:

VCO -> Panharmonium -> VCO's FM input

Monitor VCO and/or Panharmonium.

Works best with fewer voices. Blur and feedback can give interesting results too as you add/remove voices or octaves or slightly warp the frequency structure.

I was using it with FIlter 8's expo FM input, but I'm curious to try it with TZFM. It's not quite as impressive with AM though :)

User avatar
starthief
Super Deluxe Wiggler
Posts: 4407
Joined: Tue Oct 04, 2016 10:39 am
Location: St Louis, MO
Contact:

Post by starthief » Sat Aug 03, 2019 8:37 am

Here's the patch I mentioned above. Plus some background noise from E370 modulated by Kermit, and a sub drone from Rings in FM mode (which might not even be audible at this point). Valhalla delay and reverbs.

https://soundcloud.com/starthief/convergence

No sequencing -- all the pitch shifts are due to FM, and Panharmonium's delayed response to its own influence (I used a slow clock).

This was a couple of different takes mixed. Out of several takes, because finding the outlines of the sweet spot without screwing everything up was difficult.

On the other hand, when I tried stabilizing the pitch with a PLL, it was rock-solid and I could do almost anything on Panharmonium without screwing things up too much -- albeit with a certain minimum threshold of buzzy weirdness. I might have to go back to that sometime and sequence the reference oscillator.

damase
Super Deluxe Wiggler
Posts: 1790
Joined: Sun Apr 19, 2015 11:33 am
Location: Texas

Post by damase » Sat Aug 03, 2019 11:27 am

starthief wrote:This is the second time in two days I have tried to post and the post just disappeared into the ether. First time I thought it was my own mistake...

Anyway. A sound example.

http://starthief.net/stuff/panharm1.flac

DPO -> Panharmonium -> delay -> feedback into Panharmonium

Mid/side before and after Panharmonium in the feedback loop.

Just messing with knobs unscripted, with a Marbles sequence. At one point I kind of screw up and freeze Panharmonium but don't silence the DPO which is still sequenced and it's dissonant as hell.
wow, wonderful sounds here. think that seals it for me

kind of surprised at the latency, and also the surprise that its not what i would call “true stereo”. although i do think mono fx are more patchable so maybe its a good thing

User avatar
starthief
Super Deluxe Wiggler
Posts: 4407
Joined: Tue Oct 04, 2016 10:39 am
Location: St Louis, MO
Contact:

Post by starthief » Sat Aug 03, 2019 1:07 pm

FFT always introduces latency... you have to scan a certain number of samples before you can detect frequencies in it :mrgreen:

damase
Super Deluxe Wiggler
Posts: 1790
Joined: Sun Apr 19, 2015 11:33 am
Location: Texas

Post by damase » Sat Aug 03, 2019 1:19 pm

that does make sense, thank you

desolationjones
Wiggling with Experience
Posts: 452
Joined: Sat Jan 21, 2017 5:55 pm

Post by desolationjones » Sat Aug 03, 2019 2:26 pm

Have had some success syncing up the dry and wet signals using d0.

I don't know whether the Panharm architecture allows, but an internal setting for delaying the dry signal would be ace.

User avatar
starthief
Super Deluxe Wiggler
Posts: 4407
Joined: Tue Oct 04, 2016 10:39 am
Location: St Louis, MO
Contact:

Post by starthief » Sat Aug 03, 2019 11:11 pm

In the two tracks I've recorded so far with Panharmonium, I've had a lot of little glitchy click-like noises that RX6 Declick doesn't detect and required tedious manual cleanup.

With some experimentation, I managed to reproduce it in a simpler case. This is a sine in a 4-note sequence, with the clock synced to it.

Multiplier: 2
Voices: 4
Blur: about 9 o'clock
Feedback: 0
Wave: sine
Glide: 0, then I turn it up near the end.
Octave: 0
Mix: full wet

http://starthief.net/stuff/panharm_glitch.wav

There are no effects on this; the echo comes from the clock and blur. So do the glitches.

It seems like oscillators are kicking in just a tiny fraction of time before they have the proper pitch, so the "click" is the sound of them starting at the wrong frequency and sliding into place quickly. You can hear when I turn up the glide, the glitches go away.

It isn't audible on the crossfading sine, or if the blur is enough for the clock speed (faster clock seems to need more blur) or there is no blur.

User avatar
Daisuk
Super Deluxe Wiggler
Posts: 3922
Joined: Tue Jul 23, 2013 3:16 pm
Location: Oslo
Contact:

Post by Daisuk » Wed Aug 07, 2019 8:34 am

Kind of interested in this, but something tells me it probably looks a bit cooler than it sounds. Keep the demos coming. Would love to hear something a bit more melodic and less droney if someone's up for it (preferably using source material that isn't a synth). :mrgreen:

Here's a demo I stumbled upon with some interesting things. Ultimately, I don't think it's for me, but it's quite cool none the less - I just don't immediately see the musical application for it, I guess.

[video][/video]

User avatar
starthief
Super Deluxe Wiggler
Posts: 4407
Joined: Tue Oct 04, 2016 10:39 am
Location: St Louis, MO
Contact:

Post by starthief » Wed Aug 07, 2019 7:29 pm

I don't think running a full mix into it really is that useful a thing to do, even though that's how Rossum demo'd it. I tried snippets of 5 songs just this evening and... nothing good comes out IMHO.

I think you have to cherry-pick the content and dial in the filter to get a better result, and even so I'm not sure it's musically useful. Also drum loops don't sound great with it either, even in "drum mode".

It might be good with vocal only tracks, or other solo instruments, not just synths -- regardless, I think for the most part it's at its most musically useful at slower clock rates or synced, not trying to recreate as closely as possible.

I could see using it for things like following a vocal and then sending the output to a vocoder with the dry vocal (though the dry signal might require a delay to compensate for latency).

User avatar
Sinamsis
Super Deluxe Wiggler
Posts: 4316
Joined: Tue Jul 22, 2014 9:36 am
Location: Richmond, VA

Post by Sinamsis » Wed Aug 07, 2019 8:25 pm

I guess you could also run the dry signal into both the Panharmonium and something like a Fumana or even a basic envelope follower, take the Panharmonium into a VCA modulated by the envelope(s) from the Fumana to make it less droney. It's still not my cup of tea, I would be much more interested in something that generated CV rather than audio. But I could see why others might be into it.

User avatar
starthief
Super Deluxe Wiggler
Posts: 4407
Joined: Tue Oct 04, 2016 10:39 am
Location: St Louis, MO
Contact:

Post by starthief » Wed Aug 07, 2019 8:50 pm

The "droney" part is more my usage of it, both in terms of input and using a slow slice clock. It does follow input dynamics. But then that has the drawback of latency and the slice clock resolution, and the influence of blur and feedback. It's very "program dependent".


One patch I thought about and am too tired to try right now:

- Patch audio source into Panharmonium.
- Set it to 1, maybe two oscillators. Square. Other settings tuned to achieve best results.
- Patch full wet output to a PLL's "signal in" input. Set up another VCO to track with the PLL. (Homebrew PLL with XOR and slew might work here.)
- Mult the pitch CV to a notch filter.
- Run the audio source through the notch filter. Mix the Panharmonium output with that.
- Voila, you have just replaced the strongest frequency band in your audio with a squarewave, in an extremely gnarly way that might not work at all, might sound horrible, or might sound horrible in an awesome way.

I feel like this is a "galaxy brain" sort of module. It takes some doing to overcome the limitations, and it also encourages patches of types never seen before.

User avatar
Daisuk
Super Deluxe Wiggler
Posts: 3922
Joined: Tue Jul 23, 2013 3:16 pm
Location: Oslo
Contact:

Post by Daisuk » Wed Aug 07, 2019 11:21 pm

starthief wrote:I don't think running a full mix into it really is that useful a thing to do, even though that's how Rossum demo'd it. I tried snippets of 5 songs just this evening and... nothing good comes out IMHO.

I think you have to cherry-pick the content and dial in the filter to get a better result, and even so I'm not sure it's musically useful. Also drum loops don't sound great with it either, even in "drum mode".

It might be good with vocal only tracks, or other solo instruments, not just synths -- regardless, I think for the most part it's at its most musically useful at slower clock rates or synced, not trying to recreate as closely as possible.

I could see using it for things like following a vocal and then sending the output to a vocoder with the dry vocal (though the dry signal might require a delay to compensate for latency).
Thanks a lot for the info and all the demos in the thread. Really appreciate it. :)

User avatar
Jumbuktu
Super Deluxe Wiggler
Posts: 1266
Joined: Mon Mar 23, 2015 5:03 am
Location: Melbourne, Australia

Post by Jumbuktu » Wed Aug 07, 2019 11:30 pm

I am kind of surprised no one seems to have tried a patch that uses it as a sound source, playing a captured spectrum as a single voice, with modulation. Even one of the factory preset sepctra would be nice to hear. But maybe I missed that.

User avatar
Dcramer
Super Deluxe Wiggler
Posts: 5420
Joined: Fri Mar 29, 2013 12:11 pm
Location: Canada

Post by Dcramer » Sun Aug 18, 2019 11:02 am

So is it capable of letting you redefine the resynthesis sines to you own individual tunes? Or must it always use the sines to try to recreate the input harmonics? :hmm:

User avatar
starthief
Super Deluxe Wiggler
Posts: 4407
Joined: Tue Oct 04, 2016 10:39 am
Location: St Louis, MO
Contact:

Post by starthief » Sun Aug 18, 2019 12:41 pm

Dcramer wrote:So is it capable of letting you redefine the resynthesis sines to you own individual tunes? Or must it always use the sines to try to recreate the input harmonics? :hmm:
It tries to recreate the input harmonics, but you can hold down a button and turn a knob to "warp" them (different detuning per oscillator). You can sometimes get it back into harmonic territory, especially with a smaller oscillator count.

Post Reply

Return to “Eurorack Modules”