Spectral analysis for obtaining multi-band envelopes

Post your patches here! Put the format(s) in the subject.
User avatar
pugix
Super Deluxe Wiggler
Posts: 5174
Joined: Sat Oct 31, 2009 7:23 am
Location: Asheville, NC
Contact:

Spectral analysis for obtaining multi-band envelopes

Post by pugix »

Let me say what I want to do, even though the subject line hints at what I think is needed to accomplish it.

I'm interested in using feedback for making complex generated sounds. I've done a lot with patching control feedback through sample and holds to get evolving structures. Here I imagine doing a spectral analysis and extracting the instantaneous energy levels in multiple pass bands as envelopes. This looks like the first half of a vocoder. But my application is not about voice encoding. This could take as input the main audio signal output(s) from my modular. (Or it could take a miked signal of the speakers.)

Once I have eight or ten pass band envelopes, I'd use these to control any aspects of the patch making the signals/sounds. You might call this a type of spectral feedback.

Do you know of any module or circuit that would do this, without bringing the baggage of a preconceived application such as vocoding? A set of fixed, or possibly variable bandpass filters that would cover the audio spectrum from say 30 Hz to 10 KHz or so in ten bands, each having an envelope follower. I know there would be many parameters to consider here, the center frequencies, slopes, and Q of the filters, and the smoothing characteristics of the envelope followers. None of those parameters would need to be voltage-controlled, but if some of them were at least adjustable it would facilitate experimentation.

I'm looking to do this in hardware, as simply as possible.

Thanks,
User avatar
jwm
Ultra Wiggler
Posts: 905
Joined: Thu Aug 13, 2015 10:12 am
Location: florida
Contact:

Re: Spectral analysis for obtaining multi-band envelopes

Post by jwm »

sputnik spectral processor to the rescue?
https://www.analoguehaven.com/sputnikmo ... processor/

i have done very similar things with mine and it's great fun...
staaaay-shhnn
User avatar
cbm
Eardrill
Posts: 3643
Joined: Thu Dec 04, 2008 10:07 pm
Location: Alameda, California

Re: Spectral analysis for obtaining multi-band envelopes

Post by cbm »

This is a major feature of the Buchla 296 Spectral Processor, so anything inspired by that, like the Sputnik or Verbos, should get you there.
Chris Muir
Eardrill Archive - http://www.eardrill.com (on permanent vacation)
Eurorack System - Main Cabinets and Skiff
User avatar
MARK27
Ultra Wiggler
Posts: 809
Joined: Sat Mar 25, 2017 7:07 pm
Location: Oklahoma City

Re: Spectral analysis for obtaining multi-band envelopes

Post by MARK27 »

I don't think any of the Buchla-derived spectral processors are as full-featured and robust as the Frap Tools Fumana.

It's pricey, but its versatility makes it worth every penny. I was actually in the market for a fixed filter bank and I ended up springing for this.

Image
FS/T: AION 904A, Doepfer A-110-2 VCO, Erica Bl. Modulator2, Erica Dual VCF, EMW VC-104, MI Veils, Vermona twinCussion, Synthrotek Seq-8, FOH Plague Bearer, Stochastic SIG -- For prices/offers, please send me a DM

Oh god, what have I done?

My personal good BST transactions list
User avatar
Buttons ARE toys
Veteran Wiggler
Posts: 667
Joined: Wed Sep 14, 2016 9:24 am
Location: Earth

Re: Spectral analysis for obtaining multi-band envelopes

Post by Buttons ARE toys »

I've patched similar things to what you're talking about with an ADDAC601. It's only 8 bands, but each has voltage control over gain, individual outputs, and envelope followers. I especially like using it in delay feedback systems where the different envelope outputs do things like controlling compression and changing delay times to help regulate things when a particular frequency band starts to get out of control.
Astronomical Dusk - a web radio show of synthetic delights and collected oddities. Airs semi-monthly at astronomical dusk.
User avatar
pugix
Super Deluxe Wiggler
Posts: 5174
Joined: Sat Oct 31, 2009 7:23 am
Location: Asheville, NC
Contact:

Re: Spectral analysis for obtaining multi-band envelopes

Post by pugix »

All of these (Sputnik, Verbos, and Frap Tools) are too fully featured for what I need. They all do envelope extraction in multiple bands, but also have band CV inputs for vocoding and many signal outputs for complex filtering. Thanks for pointing me in the right direction. I'll keep looking.
User avatar
pugix
Super Deluxe Wiggler
Posts: 5174
Joined: Sat Oct 31, 2009 7:23 am
Location: Asheville, NC
Contact:

Re: Spectral analysis for obtaining multi-band envelopes

Post by pugix »

Buttons ARE toys wrote: Mon Jul 19, 2021 4:38 pm I've patched similar things to what you're talking about with an ADDAC601. It's only 8 bands, but each has voltage control over gain, individual outputs, and envelope followers. I especially like using it in delay feedback systems where the different envelope outputs do things like controlling compression and changing delay times to help regulate things when a particular frequency band starts to get out of control.
Getting closer. Still more features than I need, but smaller and less expensive. Thanks.
User avatar
Buttons ARE toys
Veteran Wiggler
Posts: 667
Joined: Wed Sep 14, 2016 9:24 am
Location: Earth

Re: Spectral analysis for obtaining multi-band envelopes

Post by Buttons ARE toys »

One thing I'll add is, I only ever end up using 2 - 4 band outputs for meaningful modulation in a patch. The Doepfer filterbank is super cheap on the used market, and the individual outputs mod can be built relatively easily and inexpensively. That along with a few envelope followers might do the trick while still costing less than some of the filterbanks out there. Would take up considerable space though.
Astronomical Dusk - a web radio show of synthetic delights and collected oddities. Airs semi-monthly at astronomical dusk.
User avatar
pugix
Super Deluxe Wiggler
Posts: 5174
Joined: Sat Oct 31, 2009 7:23 am
Location: Asheville, NC
Contact:

Re: Spectral analysis for obtaining multi-band envelopes

Post by pugix »

By searching modulargrid by Filter and Envelope Follower I found all of the modules already mentioned, plus the 4ms Spectral Multiband Resonator. The Spectral Multiband Resonator has six bands to the ADDAC601's eight, but has more functionality and is comparable in price.
thewrongguy
Learning to Wiggle
Posts: 4
Joined: Thu Jan 25, 2018 5:03 pm
Location: Winnipeg

Re: Spectral analysis for obtaining multi-band envelopes

Post by thewrongguy »

Never turn your back on nonlinearcircuits.

NLC Multiband Distortion
User avatar
pugix
Super Deluxe Wiggler
Posts: 5174
Joined: Sat Oct 31, 2009 7:23 am
Location: Asheville, NC
Contact:

Re: Spectral analysis for obtaining multi-band envelopes

Post by pugix »

thewrongguy wrote: Tue Jul 20, 2021 11:08 am Never turn your back on nonlinearcircuits.

NLC Multiband Distortion
Thanks, but I need envelope followers.
User avatar
ATW
Super Deluxe Wiggler
Posts: 2050
Joined: Mon Dec 02, 2019 5:01 pm
Location: Northampton, MA
Contact:

Re: Spectral analysis for obtaining multi-band envelopes

Post by ATW »

FXDf offers 6 outputs at fixed bands. One option would be to pair that with something like the A-135-4C (quad envelope follower). The cost of this would be fairly low, like $250 based on MSRP for the two. (Though the FXDf is out of production)
ᴘᴀᴛᴄʜ ᴅᴇᴄᴋ ɪs ᴀ ᴄᴏʟʟᴇᴄᴛɪᴏɴ ᴏғ ᴍᴏᴅᴜʟᴀʀ sʏɴᴛʜ ᴘᴀᴛᴄʜ ᴛɪᴘs
User avatar
MARK27
Ultra Wiggler
Posts: 809
Joined: Sat Mar 25, 2017 7:07 pm
Location: Oklahoma City

Re: Spectral analysis for obtaining multi-band envelopes

Post by MARK27 »

Hey pugix

Just for my own edification, I'd be curious to know how you end up achieving these functions you're looking for. If you do end up patching it together with discrete modules or if you go with the 4MS SMR.
FS/T: AION 904A, Doepfer A-110-2 VCO, Erica Bl. Modulator2, Erica Dual VCF, EMW VC-104, MI Veils, Vermona twinCussion, Synthrotek Seq-8, FOH Plague Bearer, Stochastic SIG -- For prices/offers, please send me a DM

Oh god, what have I done?

My personal good BST transactions list
User avatar
pugix
Super Deluxe Wiggler
Posts: 5174
Joined: Sat Oct 31, 2009 7:23 am
Location: Asheville, NC
Contact:

Re: Spectral analysis for obtaining multi-band envelopes

Post by pugix »

ATW wrote: Tue Jul 20, 2021 11:50 am FXDf offers 6 outputs at fixed bands. One option would be to pair that with something like the A-135-4C (quad envelope follower). The cost of this would be fairly low, like $250 based on MSRP for the two. (Though the FXDf is out of production)
If I could find an FXDf, I could match it up with six Universal Slope Generators, since I have twelve of them (six pairs of different makes).
User avatar
pugix
Super Deluxe Wiggler
Posts: 5174
Joined: Sat Oct 31, 2009 7:23 am
Location: Asheville, NC
Contact:

Re: Spectral analysis for obtaining multi-band envelopes

Post by pugix »

MARK27 wrote: Tue Jul 20, 2021 11:54 am Just for my own edification, I'd be curious to know how you end up achieving these functions you're looking for. If you do end up patching it together with discrete modules or if you go with the 4MS SMR.
See my last reply about FXDf and USGs as envelope followers. Use of other separate modules looks too costly.

The single module solution I like best so far is the ADDAC601. I downloaded the SMR manual and wow, it is complicated!
User avatar
WarpHead
Veteran Wiggler
Posts: 735
Joined: Fri Oct 07, 2016 7:15 am
Location: Europe

Re: Spectral analysis for obtaining multi-band envelopes

Post by WarpHead »

I may (most definitely have) read your question in the wrong way, but I'm confused.

Does your query ultimately just boil down to the cheapest/most cost-effective combination of a filterbank and envelope followers?

If so, you'd be hard-pressed to find more bang for your buck than the usual suspects in the Doepfer assortment, as mentioned.
Don't believe the hype.
User avatar
ATW
Super Deluxe Wiggler
Posts: 2050
Joined: Mon Dec 02, 2019 5:01 pm
Location: Northampton, MA
Contact:

Re: Spectral analysis for obtaining multi-band envelopes

Post by ATW »

EMW’s 2020 version of their fixed filter bank appears to have 8 bandpass outputs, along with some knob control over the bands.

Image

https://www.modulargrid.net/e/emw-fixed ... 20-version
ᴘᴀᴛᴄʜ ᴅᴇᴄᴋ ɪs ᴀ ᴄᴏʟʟᴇᴄᴛɪᴏɴ ᴏғ ᴍᴏᴅᴜʟᴀʀ sʏɴᴛʜ ᴘᴀᴛᴄʜ ᴛɪᴘs
User avatar
Hovercraft
Super Deluxe Wiggler
Posts: 1619
Joined: Fri Nov 27, 2015 12:22 am
Location: DC

Re: Spectral analysis for obtaining multi-band envelopes

Post by Hovercraft »

If you're looking for a fairly simple solution, you might consider the LZX Sensory Translator. Although it's intended for making video synth patches audio reactive, it can easily be used for straight audio patching. It has a flexible input stage for modular/line level, or a mic--and you can eq the input which is super useful. It even has a built-in mic. Three levels of slew over the envelopes and audio outputs for each pass band. It does a great job of extracting meaningful spectral energy signals from the input. I also have the Fumana and the SMR, and I recommend them both, but as you've surmised, the SMR is complex and takes some study.
Sensory_Translator_Frontpanel_1024px_900x900.jpg
User avatar
pugix
Super Deluxe Wiggler
Posts: 5174
Joined: Sat Oct 31, 2009 7:23 am
Location: Asheville, NC
Contact:

Re: Spectral analysis for obtaining multi-band envelopes

Post by pugix »

Hovercraft wrote: Fri Jul 23, 2021 9:19 am If you're looking for a fairly simple solution, you might consider the LZX Sensory Translator. Although it's intended for making video synth patches audio reactive, it can easily be used for straight audio patching. It has a flexible input stage for modular/line level, or a mic--and you can eq the input which is super useful. It even has a built-in mic. Three levels of slew over the envelopes and audio outputs for each pass band. It does a great job of extracting meaningful spectral energy signals from the input. I also have the Fumana and the SMR, and I recommend them both, but as you've surmised, the SMR is complex and takes some study.

Sensory_Translator_Frontpanel_1024px_900x900.jpg
Looks really good for my application. Smaller and less expensive than the ADDAC601. Five bands would be enough. Thanks!
User avatar
pugix
Super Deluxe Wiggler
Posts: 5174
Joined: Sat Oct 31, 2009 7:23 am
Location: Asheville, NC
Contact:

Re: Spectral analysis for obtaining multi-band envelopes

Post by pugix »

User avatar
Dave Peck
Super Deluxe Wiggler
Posts: 5744
Joined: Thu Mar 24, 2011 1:22 pm
Location: SF bay area

Re: Spectral analysis for obtaining multi-band envelopes

Post by Dave Peck »

There are several options for the filterbank with individual band outputs, and a few are pretty affordable so you could just go for the cheapest (for this purpose the actual audio quality doesn't matter that much, you just need to derive the envelope). But if you then have to buy a pile of envelope followers in addition to that, this method will get expensive, The cheapest price I've seen for an envelope follower module is maybe $75 for a Ladik EF.

So it may be best to buy a module that includes both the filter bank and EF functions like a Bark filter processor:
https://www.modulargrid.net/e/verbos-el ... -processor

So that really gives you two-thirds of a complete patchable vocoder system - the filterbank to separate the audio signal into separate bands, and the set of EFs to convert that audio into a set of control voltage signals.

But, according to rules of GAS, this mean you really ought to consider getting a complete patchable vocoder system so it covers what you are trying to do now, as well as all other vocoder duties. And since it will be a fully patchable vocoder, you can 'scramble' the bands and use it for LOTs of cool things beyond just making your synth talk.

So that would be something like the Verbos module plus a set of corresponding band pass filters, or something like the COTK vocoder suite being discussed in this thread: viewtopic.php?f=24&t=251196 which provides these same functions.

Or you can construct it from standard modules by getting any affordable FFB with individual band outputs (like Dotcom with the output expander), plus a pile of EGs, plus a pile of BPFs, but all those EFs and BPFs will add up.
User avatar
Snufflepuff
Veteran Wiggler
Posts: 684
Joined: Sat Dec 19, 2020 11:53 pm

Re: Spectral analysis for obtaining multi-band envelopes

Post by Snufflepuff »

“I've done a lot with patching control feedback through sample and holds to get evolving structures.“

This sounds intriguing. Could you give an example? Thanks.
User avatar
pugix
Super Deluxe Wiggler
Posts: 5174
Joined: Sat Oct 31, 2009 7:23 am
Location: Asheville, NC
Contact:

Re: Spectral analysis for obtaining multi-band envelopes

Post by pugix »

Snufflepuff wrote: Mon Aug 09, 2021 2:37 pm “I've done a lot with patching control feedback through sample and holds to get evolving structures.“

This sounds intriguing. Could you give an example? Thanks.
Have a look here.

https://pugix.com/synth/eurorack-quantussy-cells/

https://pugix.com/synth/tag/automatic-music/
Post Reply

Return to “Synthesis Techniques”