synthesizing a choir

musquash · Post by **musquash** » Tue Jan 29, 2002 6:36 am

is it possible to set up one of the standard synthesizers (inclusive propack or free devices) that it sounds like the voice of a choir?
i need something like that for the background but i couldnt figure it out yet.

paulrmartin · Post by **paulrmartin** » Tue Jan 29, 2002 6:38 am

Try the Modular2 with the Waldorf oscillators.

You have to work on the filtering a bit but you might achieve what you're after.

Good luck!

subhuman · Post by **subhuman** » Tue Jan 29, 2002 9:34 am

Also check out the Modular2's vocal filters, this can help quite a bit, and may be a little easier to get 'vocal' sounds from the Waldorf oscillators (you will need to find & use the right wavetable for this of course).

Here is an article on the web which might also help you a bit:
<a href=http://www.sospubs.co.uk/sos/mar01/arti ... sp>FORMANT SYNTHESIS</a> - luckily, the Vocal filters tend to help get the proper frequency ranges so the effect is easier to accomplish.

Hope that helps, this is a fun one!

kensuguro · Post by **kensuguro** » Tue Jan 29, 2002 10:41 am

Here's a great link
http://www.helsinki.fi/hum/hyfl/projekt ... t_eng.html
Lots of formant charts to play with.

paulrmartin · Post by **paulrmartin** » Tue Jan 29, 2002 10:58 am

WOW! My singing teacher is going to like this page!

Thanks, Ken!

Paul

kensuguro · Post by **kensuguro** » Tue Jan 29, 2002 12:31 pm

Some more strange toys if you're interested:
http://fonsg3.hum.uva.nl/paul/
THE praat analysis/synthesis engine, free, and is available for most platforms. Very high end analysis tool.
It does some VERY funky things, like totally strip a vocal of its formant data.. and stuff like that. Think about it.. it's like chopping off your head, and just listening to the glottis vibrate. Amazing! (and I was thinking of keeping this trick under my sleeves!)

So you can analyse your own formant with this, and then implement it in mod2 for your own customized choir. :> Good luck. I still haven't had the time to explore this software. It's VERY complex.

[ This Message was edited by: kensuguro on 2002-01-29 12:37 ]

musquash · Post by **musquash** » Wed Jan 30, 2002 7:19 am

making a choir is tougher than expected.. thanks for your help, i will upload my choir when it's presentable

paulrmartin · Post by **paulrmartin** » Wed Jan 30, 2002 7:23 am

Can't wait!

Spirit · Post by **Spirit** » Wed Jan 30, 2002 7:29 am

Celmo's Deep Blue (uses the Waldorf salad) can do a pretty fair voice, but you'd need lots of effects and tracks to get a choir - and this goes for any synth methinks. Why not go for samples, or have you set yourself a challenge

musquash · Post by **musquash** » Wed Jan 30, 2002 7:41 am

i prefer to avoid samples - it gives me a good feeling to know i did everything only with synthesis. actually i would be really proud if i could make my pulsar singing for me

Spirit · Post by **Spirit** » Wed Jan 30, 2002 7:42 am

'tis a noble goal you seek

kensuguro · Post by **kensuguro** » Wed Jan 30, 2002 8:58 am

I'm trying to keep myself from going into the business of vocal modeling again.. But does anyone know of sources that tell you how the glottis vibration changes as the singer changes pitch? This is right where my research stopped, cuz I can't find any information on it.
Ah well, someday, I'll continue my research. I wish you good luck! I'd be glad to share my findings about the human voice. You'd be surprised at how many linguists are interested in voice synthesis... and so little music people!!

[ This Message was edited by: kensuguro on 2002-01-30 09:00 ]

paulrmartin · Post by **paulrmartin** » Wed Jan 30, 2002 9:02 am

Hey! Does anybody remember using analog synths making the resonant filters go "ow" and "yoy"?

Michu · Post by **Michu** » Wed Jan 30, 2002 1:56 pm

THE praat analysis/synthesis engine, free, and is available for most platforms. Very high end analysis tool.
It does some VERY funky things, like totally strip a vocal of its formant data.. and stuff like that. Think about it.. it's like chopping off your head, and just listening to the glottis vibrate. Amazing! (and I was thinking of keeping this trick under my sleeves!)

Ken, this is great!!!!
thanks a lot!

[ This Message was edited by: Michu on 2002-01-30 13:57 ]

at0m · Post by **at0m** » Wed Jan 30, 2002 1:56 pm

You want a synth that sings?

I know one that can talk, http://www.neutron7.com, the "W".

at0m · Post by **at0m** » Wed Jan 30, 2002 2:03 pm

And this one, also by DeFeX!!
Robotronix

Michu · Post by **Michu** » Wed Jan 30, 2002 8:14 pm

this is rather funky stuff than choir synthesis, but anyway...

Eurocide · Post by **Eurocide** » Wed Jan 30, 2002 10:21 pm

Apoptygma Berserk: "Kathy's song (come lie next to me)"

Sung with c64-like speech synthesis on a mac.

Just my two cents...
Eurocide-HQ

at0m · Post by **at0m** » Thu Jan 31, 2002 6:53 pm

Can't really remember which one, but RadioHead has this song and one computer voice speaks/ does the lyrics. Cool, how they integrated that simple tekst reader into a song. It says the weirdest stuff just with this dry synthetical monotone voice...

kensuguro · Post by **kensuguro** » Sun Feb 10, 2002 10:43 am

Just a reminder.. In case you haven't checked out my post in Mod forum, I've found an equation that roughly estimates frequency values in MID units (0-127). You may want to use it, if you're going to do FFT analysis, and implementing that as a bank of filters in Mod2. Also, as a general guideline, as I've learned from making the vocoder, the characteristics of the human voice are concentrated in the 160-6000 zone. So, if you want to convert from FFT filters, make sure you have lots of filters in that area. And then leave a few for the sibillances. (I guess you might already know that)

[ This Message was edited by: kensuguro on 2002-02-10 10:45 ]