What exactly is a "voice"?

[Benny Hill]

Boy all kinds of sarcastic comments could be coming my way.

I've heard the term "voice" being used to describe synthesizers. But it seems like I've heard it used in different ways.

Today I heard someone say that they have a simple one voice modular system. And I've heard synthesizers being described as 2 voice etc... What makes a system one or more voices? I'm assuming one voice is an osc-filter-eg. So two voices would have 2 osc-2 filters- 2 EG's?

[djugel]

It's how many notes you can play at once.

[MadforBrad]

Independent instances of a generated sound. A 1 ossillator synth could have 100 voices nothing to do with ossiators.

[Benny Hill]

Originally Posted by djugel

It's how many notes you can play at once.

Oh geez is that's simple? Now i feel stupid.

[supracg]

Originally Posted by Benny Hill

Oh geez is that's simple? Now i feel stupid.

Multiplied by the number of voices you set the synth at, I believe

[djugel]

Originally Posted by Benny Hill

Oh geez is that's simple? Now i feel stupid.

Nah,.. it's actually less stupid because you asked the question.

[Kindred]

its not a stupid question at all - I remember ages ago looking for synth glossaries which clearly explained it and there are not many. When you are starting out, the whole polyphony/voice/multi-timbral thing is very confusing - it took me ages to work out the difference! it used to be more relevant when everything was outboard - most vsti's are not particularly limited in these aspects

[Yoozer]

Originally Posted by Benny Hill

So two voices would have 2 osc-2 filters- 2 EG's?

On an analog/virtual analog synth: no.

On a sample-based synth or sampler: yes.

The JV1080 has 64 "voices" of polyphony. What they don't mention is that up to 4 voices can be stacked on top of eachother in a preset, so if you want to use two "oscillators", the effective polyphony is halved to 32 notes. Each sample-based oscillator has its own chain of LFOs, filters and envelopes attached to it. On (virtual) analogs, those oscillators are first mixed before they go into a singular filter.

To complicate even more: the Access Virus (B and C) have 24 and 32 notes of polyphony, but if you enable the third oscillator this is cut to 16 and 24 respectively. The new Virus TI has dynamic allocation, which means that polyphony depends on how complex a single voice is - oscillator type used, number of oscillators enabled, any (audio rate) modulation.

When someone's talking about a modular, there's a degree of flexibility. For instance, several monosynths offer 2 oscillators, 1 filter + envelope, 1 amplifier + envelope and 1 LFO. For each of those, you'd need to buy modules. If you buy twice the amount, you get 4 oscillators, 2 filters + 2 envelopes, 2 amplifiers + 2 envelopes and 2 LFOs. You could use all 4 oscillators at the same time and get a much more complex single voice, or you could use them as two separate voices with a simplified structure.

I can detune two oscillators so one plays the root and the other plays the fifth. That's not two voices of polyphony, because the detuned oscillator is slaved to the first. There are a lot of tricks possible here if you use some really clever modulation - including having it play a complete melody at a different time - but the control signal and the way they go into the mixer would determine whether it's a real separate voice or not.

Originally Posted by Kindred

When you are starting out, the whole polyphony/voice/multi-timbral thing is very confusing - it took me ages to work out the difference! it used to be more relevant when everything was outboard - most vsti's are not particularly limited in these aspects

Exactly. There's no real difference between multitimbrality and sending incoming MIDI notes to several plugins at once anyway.

16-part multitimbrality means that you get 16 distinct synthesizers with the caveat that they have to share polyphony and effects. So, for that JV1080: if you're playing a 3-note chord on the piano which requires two of those voices per note, you have 64 - (3 * 2) = 58 notes left for all the other instruments. Earlier multitimbral instruments can be very rigid in this regard in the sense that they want you to define the number of used notes per channel in advance, instead of allocating it dynamically.

[Apprendista]

Originally Posted by Yoozer

On an analog/virtual analog synth: no.

On a sample-based synth or sampler: yes.

Thinko? Of course a 2-voice synth has two identical signal paths, oscillators, filters and all, analog or not.

[gremlin moon]

A lot of people like being in the box because your voice limitations (how many sounds you can play at once) with a modern computer are pretty immense.

The classic "purist" debate tends to center around the issue of sound quality:
"I'd rather individually track my monophonic analog 30 times because it sounds better than any VST"

And workflow:
"You can get to more interesting places with hardwired knobs and no screen and no mouse" vs "with my laptop I can be open to creativity anywhere and I can assign knobs to whatever I want with a midi controller and have fairly unlimited instances of a synth."

I've noticed a lot less purist debates as of late -- at the end of the day slutz aren't picky and they like variety

[ofajen]

Originally Posted by Yoozer

Earlier multitimbral instruments can be very rigid in this regard in the sense that they want you to define the number of used notes per channel in advance, instead of allocating it dynamically.

As my electronic music prof pointed out to me many years ago, there are plusses and minuses to both static and dynamic allocation. For instance, say you want a synth that can play eight simultaneous notes to do a bass sound, a mono lead and some rhythmic chord parts with up to three note chords sounding and a amplitude envelope with a bit of release length to it. If you have all those parts going on at the same time and dynamic allocation of voices, the sustained bass note may get jettisoned long before the note is supposed to end, due to all the changing, accumulating notes in the chord part. Static allocation in that case means you won't be left with a big hole in the music. In theory, "first note" allocation priority should cover this, but it doesn't always seem to work out that way.

Cheers,

Otto

[enossified]

Originally Posted by ofajen

As my electronic music prof pointed out to me many years ago, there are plusses and minuses to both static and dynamic allocation.

It's simple:

Static allocation wastes voices if each part is not using all the notes it has been allocated all the time. This was a real big deal back when you had multitimbtral synths with only 6-8 voices. I suppose on a modern synth with 128 voices it would be a little easier to handle

The Ensoniq ESQ-1 was the first synth to offer dynamic allocation and stretched it's 8 voices a lot farther. Yes, you do have to worry about voices cutting off but that just means you need to be more clever when sequencing.

Modern instruments have very complex allocation algorithms that take into account things like velocity switches, key ranges, etc. to maximize the pool of free voices.

[Yoozer]

Originally Posted by Apprendista

Of course a 2-voice synth has two identical signal paths, oscillators, filters and all, analog or not.

If they're truly polyphonic. Paraphonic will still count voices as separate notes that you can play and trigger at leisure, but you're just going to get a single filter.