Vocal qualities into a synth patch

[FarleyCZ]

Hi everyone. It's not completely u- he topic, but as I'd like to achieve this in Zebra, here is the right place to ask.

I've been thinking lately. Is there way, how to make lead synth patch equally enjoyable as human vocal? Is there even way, how to make them sound similar? With few hours, everyone can do some wierd formant female aah or monk sound ... But what about male vocals? ...or any other type really.

I know this is hard topic, touching physical modeling like crazy, but anyway. What's all the stuff I should do on a synth to "vocaize" it as much as plssible?

I guess first whats comming to mind is formant information. There is first problem I found. In all formant filters, there is usually just two formant peaks. I have no idea how other peaks that (acording to wiki) human vocal chors has behaves... Any help?

Pitch is also a question. Some amount of portamento and vibrato is a must.

...but whats the rest? I found by analyzing vocals, pianos and other real instruments that they have just a small amount of of fundamental frequency. But how to get so smooth, complex and nice harmonics, that they sound good by themselfs? Everytime I get rid of fundamental in my leads, everything sounds bad. It can't be just question of filters, they take stuff away, but I need it there, just in diffeeent 'configuration'.

Any ideas?

[3ee]

Instead of formant filters, try to build the partials using the osc in additive mode + carefully select which oscfx to use... cause... there are only 2 slots
Z3 will have 3

or maybe FM, but with resonant filters in the way, that could work maybe?!

[pdxindy]

Instead of formant filters, try to build the partials using the osc in additive mode + carefully select which oscfx to use... cause... there are only 2 slots
Z3 will have 3

or maybe FM, but with resonant filters in the way, that could work maybe?!

Please lets not start talking about Z3 and what will be in it. It is only in the planning stage and we do not know what it will finally be... also, it is like a year away at best and could easily be longer...

[pdxindy]

Hi everyone. It's not completely u- he topic, but as I'd like to achieve this in Zebra, here is the right place to ask.

I've been thinking lately. Is there way, how to make lead synth patch equally enjoyable as human vocal? Is there even way, how to make them sound similar? With few hours, everyone can do some wierd formant female aah or monk sound ... But what about male vocals? ...or any other type really.
Any ideas?

Not really sure what you are looking for... the very best synths dedicated to vocal sounds do not come close to the human voice and its remarkable expressiveness...

[stillshaded]

There's a lot of things that would go into it, I'll throw a little something into the stew.


Vibrato and portamento for sure, however.. IMO you will have to automate the rate of both heavily to achieve the natural quality of the human voice. Often the vibrato will start fast and speed up, and a singer will glide between different notes at different speeds without even thinking about it. Something to consider.

[FarleyCZ]

Thanks stillshaded! That's exactly the kind of thinking I'm talking about! I know we can't reproduce vocals completely, but why not to think about it at least?

...or any real instrument really. Except combinator-ish guitar, tbh I struggle to make things "lifelike" more as I'm going down the scale ... Always that harmonics vs. fundamental problem I described earlier ... It's also the reason why every try of male vocal sound like a monk ... Those harmonics just need to be generated somehow differentaly...

[Howard]

Apart from all the above, you can get strong "vocal" colouration by putting an Allpass filter in parallel with the main signal (in the FX grid is often good enough). One of my fave tricks.

[trance_lucent]

Apart from all the above, you can get strong "vocal" colouration by putting an Allpass filter in parallel with the main signal (in the FX grid is often good enough). One of my fave tricks.

Analyzing new The Zebra Orchestra soundset I can often see this trick too - it seems allpass filter somehow makes soond more "organic" and less "synthetic".

[pdxindy]

Apart from all the above, you can get strong "vocal" colouration by putting an Allpass filter in parallel with the main signal (in the FX grid is often good enough). One of my fave tricks.

+1

I love the allpass filter!

[pdxindy]

Thanks stillshaded! That's exactly the kind of thinking I'm talking about! I know we can't reproduce vocals completely, but why not to think about it at least?

...or any real instrument really. Except combinator-ish guitar, tbh I struggle to make things "lifelike" more as I'm going down the scale ... Always that harmonics vs. fundamental problem I described earlier ... It's also the reason why every try of male vocal sound like a monk ... Those harmonics just need to be generated somehow differentaly...

a keyboard is not well suited to vocal sounds... a wind controller would be better... or at least a breath controller for the keyboard...

[FarleyCZ]

Must admit I didn't explore allpass filter too much. Thanks for the tip!

...another detail I've noticed. Voice can kind of morph between octaves a little without changing it's harmonical content too much. How can I model that? Adding sub-sine? ...dunno. Two harmonically rich oscilators are nonsense, it creates kind of "nasty" stuff as higher harmonics try to blend together. On the other side one proper OSC and subsine makes unnaturally high sound.

[dmbaer]

Apart from all the above, you can get strong "vocal" colouration by putting an Allpass filter in parallel with the main signal (in the FX grid is often good enough). One of my fave tricks.

Very interesting, Howard. Could you elaborate just a little? For example, where do you typically place the cutoff (high, mid or low)? Do you modulate the cutoff? If the signals are parallel, do you typically match the levels, or do you modulate the relative balance, or something else?

[Sendy]

A lot of vocal expressiveness comes from paying attention to pitch, portamento and vibrato. Having a fixed time portamento and fixed vibrato isn't going to cut it. Neither is having all your pitches bang on the 12-tet grid (finetune some key notes until they 'pop').

Formants created by filters or perhaps sync or other effects should not keytrack significantly. Complex sync networks can produce vocal-like sounds. Often, resonant peaks move in contrary motion in the production of vowel sounds.

Harmor lets you load in a wave like someone speaking and play it on any key and scrub through it. You can get amazing vocal stuff that way.