FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 

Choir Synthesizer patch
MUFF WIGGLER Forum Index -> Production Techniques Goto page 1, 2  Next [all]
Author Choir Synthesizer patch
OK, so I am interested in trying to make an analog "choir" type sound with my modular.

I used to have the Korg Lambda ES-50 and it had a sweet (cheezy) chorus/vocal sound, which I have occasionally missed, and it seems like I should be able to pull off something like this.

I have a Doepfer FFB, an MOTM e340, reverbs, plenty of modules... so it seems like something within reach, but my efforts so far sound crappy. I don't need realistic, but I want that airy, almost misty "ah" sound.

Advice anyone?
Ken Elhardt?
Mike Peake?
I don't have the Lambda to reference. That era of gear usually had ensemble/chorus units containing several delays that aren't yet very easy or cheap to duplicate given the usual need for modulation by LFOs going the same rate but in different phases. Generally I think that's a major problem with a modular for this kind of use. It's got to cost way more than a used lambda to nail. Just the quality of the overlapping notes. Reverb won't do that convincingly to a mono sound. You are hearing layered multiple polyphonic voices decaying into eachother.

The FFB will definitely help. Again, I'm not an expert on that ensemble kbd but generally you'd have a fairly simple organ sound being fixed filtered to give it a character and being an organ, you could add octaves to create the character of a stop. I'm guessing the e340 won't be that useful since the original didn't have anything roughly similar going on inside. I don't think the sound comes from a lot of simple sounds detuned and moving. You have a generally simple sound with likely a strong pitch vibrato going through a very characterful ensemble with another layer of a different keyboardy sound that's probably what you aren't after.
Interesting... that makes sense that the ensemble unit had a lot to do with the sound.

Any recommendations of chorus pedals or modules to get me closer to that sound?

Is it possible to do "pretty" chorus with Doepfer BBDs?
Seems like the cheapest, best option would be to patch out to a proper chorus FX unit, perhaps rack or stompbox... best I've heard personally is still the Diamond Halo.

...but I'm not so sure it does "vocal" sounding chorus...
Pretty as in expensive with doepfer. Ensemble keyboards tend to have a triple delays so one chorus unit isn't going to do it imho. One would probably come closer with digital simulations.

The vocal character generally filtering. I'm not sure what they are doing but most of the pros use 3 or 4 resonant band pass filters. You can find the frequencies for different vowels all over the net. Korg would surely be saving money not using VC filters or a set per voice.

Though some of the character is surely the long decaying polyphonic tones of the last note still dying away while the next one plays.
i got some simple and effective results from a regular subtractive polyphonic patch. rich oscillators with resonating low passes a little above root notes to make 'ahh' then fed into long reverb.
Lots of ideas to try as soon as I get everything plugged back in!
(moving gear to a different room)

Keep the ideas coming, please.
I use a Casio SK-1 (with me going 'ahhh' or 'mmmm') split into a couple of effects chains then add some reverb/delay/etc. in the mixer w/ effects sends. Seems like a lot of work when you read it, but all the stuff is here already. It is amazing how often I use my SK-1 for something.
I've been able to make a pretty cool choir patch on the Nord Modular G2. The basic idea was to feed the oscillator through four different formant filters in parallel. Each formant filter set to a different pair of vowel sounds, and continuously morphing between the two. The output of each formant filter is run through a VCA with an LFO controlling its amplitude, and then all four are mixed together for final output. The net result is a constantly evolving choir with different elements coming and going at varying rates.

I've had great success reproducing the morphing formant filter effect with the Doepfer A-127 triple resonant filter controlled by an A-155 sequencer. Of course, to reproduce my G2 patch, you'd need four of those filters...
I had a you tube or something I found about how to do this, plus tables for the frquency to set bandpass filters to for the vowels..can't find it now. It's out there..somewhere.
chamomileshark wrote:
I had a you tube or something I found about how to do this, plus tables for the frquency to set bandpass filters to for the vowels..can't find it now. It's out there..somewhere.
Just paid for a used MOTM 410, which I hope will assist me in my search for vocal timbres.
Bandpass Filter Frequencies for Formant Synthesis

From Csound:

Soprano 'a'      f1   f2   f3   f4   f5
frequency (Hz)   800 1150 2900 3900 4950
Amp (dB)         0    -6  -32  -20  -50
bw (Hz)          80   90  120  130  140

Soprano 'e'      f1   f2   f3   f4   f5
frequency (Hz)   350 2000 2800 3600 4950
Amp (dB)         0   -20  -15  -40  -56
bw (Hz)          60  100  120  150  200

Soprano 'i'      f1   f2   f3   f4   f5
frequency (Hz)   270 2140 2950 3900 4950
Amp (dB)         0   -12  -26  -26  -44
bw (Hz)          60   90  100  120  120

Soprano 'o'      f1   f2   f3   f4   f5
frequency (Hz)   450  800 2830 3800 4950
Amp (dB)         0   -11  -22  -22  -50
bw (Hz)          70   80  100  130  135

Soprano 'u'      f1   f2   f3   f4   f5
frequency (Hz)   325  700 2700 3800 4950
Amp (dB)         0   -16  -35  -40  -60
bw (Hz)          50   60  170  180  200

Alto 'a'         f1   f2   f3   f4   f5
frequency (Hz)   800 1150 2800 3500 4950
Amp (dB)         0    -4  -20  -36  -60
bw (Hz)          80   90  120  130  140

Alto 'e'         f1   f2   f3   f4   f5
frequency (Hz)   400 1600 2700 3300 4950
Amp (dB)         0   -24  -30  -35  -60
bw (Hz)          60   80  120  150  200

Alto 'i'         f1   f2   f3   f4   f5
frequency (Hz)   350 1700 2700 3700 4950
Amp (dB)         0   -20  -30  -36  -60
bw (Hz)          50  100  120  150  200

Alto 'o'         f1   f2   f3   f4   f5
frequency (Hz)   450  800 2830 3500 4950
Amp (dB)         0    -9  -16  -28  -55
bw (Hz)          70   80  100  130  135

Alto 'u'         f1   f2   f3   f4   f5
frequency (Hz)   325  700 2530 3500 4950
Amp (dB)         0   -12  -30  -40  -64
bw (Hz)          50   60  170  180  200

CounterTenor 'a' f1   f2   f3   f4   f5
frequency (Hz)   660 1120 2750 3000 3350
Amp (dB)         0    -6  -23  -24  -38
bw (Hz)          80   90  120  130  140

CounterTenor 'e' f1   f2   f3   f4   f5
frequency (Hz)   440 1800 2700 3000 3300
Amp (dB)         0   -14  -18  -20  -20
bw (Hz)          70   80  100  120  120

CounterTenor 'i' f1   f2   f3   f4   f5
frequency (Hz)   270 1850 2900 3350 3590
Amp (dB)         0   -24  -24  -36  -36
bw (Hz)          40   90  100  120  120

CounterTenor 'o' f1   f2   f3   f4   f5
frequency (Hz)   430  820 2700 3000 3300
Amp (dB)         0   -10  -26  -22  -34
bw (Hz)          40   80  100  120  120

CounterTenor 'u' f1   f2   f3   f4   f5
frequency (Hz)   370  630 2750 3000 3400
Amp (dB)         0   -10  -26  -22  -34
bw (Hz)          40   60  100  120  120

Tenor 'a'        f1   f2   f3   f4   f5
frequency (Hz)   650 1080 2650 2900 3250
Amp (dB)         0    -6   -7   -8  -22
bw (Hz)          40   60  100  120  120

Tenor 'e'        f1   f2   f3   f4   f5
frequency (Hz)   400 1700 2600 3200 3580
Amp (dB)         0   -14  -12  -14  -20
bw (Hz)          70   80  100  120  120

Tenor 'i'        f1   f2   f3   f4   f5
frequency (Hz)   290 1870 2800 3250 3540
Amp (dB)         0   -15  -18  -20  -30
bw (Hz)          40   90  100  120  120

Tenor 'o'        f1   f2   f3   f4   f5
frequency (Hz)   400  800 2600 2800 3000
Amp (dB)         0   -10  -12  -12  -26
bw (Hz)          40   80   100  120  120

Tenor 'u'        f1   f2   f3   f4   f5
frequency (Hz)   350  600 2700 2900 3300
Amp (dB)         0   -20  -17  -14  -26
bw (Hz)          40   60   100  120  120

Bass 'a'         f1   f2   f3   f4   f5
frequency (Hz)   600 1040 2250 2450 2750
Amp (dB)         0    -7   -9   -9  -20
bw (Hz)          60   70   110  120  130

Bass 'e'         f1   f2   f3   f4   f5
frequency (Hz)   400 1620 2400 2800 3100
Amp (dB)         0   -12   -9  -12  -18
bw (Hz)          40   80   100  120  120

Bass 'i'         f1   f2   f3   f4   f5
frequency (Hz)   250 1750 2600 3050 3340
Amp (dB)         0   -30  -16  -22  -28
bw (Hz)          60   90   100  120  120

Bass 'o'         f1   f2   f3   f4   f5
frequency (Hz)   400  750 2400 2600 2900
Amp (dB)         0   -11  -21  -20  -40
bw (Hz)          40   80   100  120  120

Bass 'u'         f1   f2   f3   f4   f5
frequency (Hz)   350  600 2400 2675 2950
Amp (dB)         0   -20  -32  -28  -36
bw (Hz)          40   80   100  120  120

Any tips on getting my various filters set to those actual frequencies?
I don't have a frequency counter... is there any software that does this conveniently?
I've usually got good results just doing it by ear. Best thing to start off w/is a pulse wave, tune the mark/space ratio till it sounds kind of vocally. I mix a couple of bandpass filters, a lowpass and a highpass. I would start with the 2 bandpasses, tune them till it sounds like and "aah" sound, then use very small amounts of the lowpass and highpass to fill out the bottom end and add a bit of air. I made a really nice set of samples using these techniques, feeding my ES50 Lambda through a Serge filter network and a Frostwave resonator. I'll see if I can scare the samples up and upload them. The Serge variable bandwidth filter is VERY good for this, BTW.

there is another technique, someone years ago posted it on the ~AH list, I couldn't find a link quickly. this involves a pair of oscs, one running through a filter, the second one hard syncved to the first, not running through the filter but modulating it at audio rate, and tuned quite a bit higher than the audio osc. I tried this when it was posted, and got some quite striking results. It's probably woth experimenting with that patch. If I can find the post, I'll link to it.
rafe127 wrote:
Any tips on getting my various filters set to those actual frequencies?
I don't have a frequency counter... is there any software that does this conveniently?
I use a Tiptop oscillator with frequency counter, but all you need is to find a simple tone generator for your computer. Start generating the first frequency, turn up filter resonance, and sweep the cutoff. You'll easily hear it when you hit the right spot. Repeat...
i managed to patch a convincing male "ah" sound from the sos article the other day, with three band pass filters in parallel.

what deastman said is how i did it: turn up res on all filters, use the pc to generate sine tones, and tune to unison/no beating.

i then got all crazy by modulating the three filters very slightly with different lfos for morphing vowel sounds.
i've spent a really long time trying to emulate human vowels...

the best results I've had are from simple subtractive patches with 1 or 2 filters, but with really really careful attention to detail. It's really subtle things that make something sound convincing or not. A touch of modulation, really precise filter placement and movement.

I rarely got good results from using the charts. By ear is totally the way forward imo, and much more fun/expressive/alive.

Camel Audio 'Alchemy' is fascinating to study hamronic structure in voices, but it's a really tough method to take it beyond a sample and into an alive sound.

For my own music now I think I'm just aiming for nice sounds and methods in the near immediate moment, rather than aiming specifically for human sounds from preplanned sound design.
bumping this thread as I am trying to do a women's choir patch. there are some good ideas here but let's hear some more.
The A-127 is indeed good for this. The manual has example vowel frequencies to resonate (ie bandpass filter) listed, as does the Csound manual as noted above. (The Csound manual also has other useful tables, such as ISO 16 standard pitches.)

If you want to get really into this, I'd recommend chapter 5.7 of Hermann Helmholtz's The Sensations of Tone, Vowel Qualities of Tone.

If you just want a nice choir sound, I'd go with a sample CD of a Mellotron or Orchestron, or pay a session singer to sing scales for me. Personally, I'd recommend Mellotron Archives, Optigan Orchestron Talentmaker and Ann Bailey for this. Incidentally, if you take a fairly high note sung by a female singer and pitch it down an octave or two, you end up with a new age pad.

Good luck, and let as all know how it goes!
Thanks ZoeB! I am checking out that chapter. I haven't had a chance to try something yet. I don't want to use samples. I am determined to get something decent with the modular. If it turns out well, I will post a clip.
mojopin wrote:
Thanks ZoeB! I am checking out that chapter. I haven't had a chance to try something yet. I don't want to use samples. I am determined to get something decent with the modular. If it turns out well, I will post a clip.


Well in that case, I'd recommend wiring up two or three bandpass filters in parallel (which is what the A-127's great for, but at a pinch, you could probably use any combination of bandpass filters and a mixer), and using your DAW's built-in spectrum analyser (for example, Reaper's ReaEQ is good for this), twiddle the cutoff points until they match the ones in that chapter, or the A-127's manual, or the Synth Secrets issue on the topic.

Good luck! ^.^
Yup, that's the plan. I can get five bandpasses out of my setup (plus i have the fun dual peak of the mmf-1).

I have been reading academic papers about choirs and one thing I found was this:
The so-called ‘chorus’ or ‘ensemble’ effect is interest-ing both musically and perceptually. It is usually imitated in effect devices using slowly varying time shifts, giving the impression of rotating speakers rather than of an en-semble. Dolson found in 1983 that the quasi-random am-plitude modula-tion of beating partials alone can cue the perception of ensemble. The small changes in frequency, he found, are less salient perceptually. This suggests an alternative simulation of the chorus effect.

So I plan on sending small amounts of chaotic and random cv to the cutoff and amplitude of each band-pass (finally a use for the JAG!). The paper's authors go on to try it and don't find success but it is a bit hard for me to follow their methods and terminology. Anyhow, I wouldn't have thought about modulating the bp's amplitude.

I think it helps that modulars are generally monophonic, as are voices, so you won't be tempted to play block chords, which are the main thing. When orchestrating any kind of orchestral instruments, including choirs, the main thing is to treat it as a bunch of monophonic instruments, not one polyphonic one. So you can play your lowest notes in one pass, medium ones in the next one, and so on. That means you can also change the patch from one range to the next, so, for instance, the higher the range, the more swift the patch is in terms of envelopes and portamentos. You can also simply record several takes of each note, especially again slightly tweaking the odd parameter here and there in between, and/or during, each performance. I've made very basic stabs at wiring up string patches, rather than choirs, so far, and although they didn't sound anything like strings (which I didn't really intend them to -- I have samples for that), they did start to behave a bit like a real orchestra. So yeah, embrace multitracking a monophonic instrument!

Oh, and it also means you can have separate portamento for each monophonic line, all happening at once, so some voices are going slowly down, some slowly up, and some quickly up, all at the same time. And you can embellish each part with the odd grace note and such, to really get out of block chord territory. That makes a big difference.
MUFF WIGGLER Forum Index -> Production Techniques Goto page 1, 2  Next [all]
Page 1 of 2
Powered by phpBB © phpBB Group