VST Output levels

[cixelsyd]

So I am in the middle of testing a small synth I have developed, and I'm referencing my output gain with other synths. One thing I'm noticing is that if I play a single note my gain level is fairly equal to other synths measuring with the same VU meter. However if I play multiple notes my gain level increases while other commercial synths decrease in gain.

I'm wondering if its common to put a compressor/limiter on the main out and if this is indeed what I am seeing in the commercial synths.

[BertKoor]

If you add uncorrelated signals, the VU meter should show +3dB for each doubling of voices.
I think it's bad practice to put a compressor/limiter in a VSTi. You should be able to figure that out by lowering the output level with 20dB or so.
Leave the user totally in control of the volume. There are gain knobs everywhere, store the corrected volume with the patch.

[BertKoor]

If you add uncorrelated signals, the VU meter should show +3dB for each doubling of voices. It's +6 for correlated signals. So in a synth context maybe +4 think4.5.
I think it's bad practice to put a compressor/limiter in a VSTi. You should be able to figure that out by lowering the output level with 20dB or so.
Leave the user totally in control of the volume. There are gain knobs everywhere, store the corrected volume with the patch.

[cixelsyd]

Hi BertKoor, and thanks for the response. What you are saying is basically what I think I'm seeing/hearing. I appreciate your comment. I was reluctant to put a compressor at the main out simply because I kinda like th e rawness of the sound. Compression, makes it sound kinda 'smoother'.

Thanks again for quick post. Can I say after finishing a fairly basic synth I now have a huge respect for people who do this everyday. I feel like I've climbed everest twice.

All my best

[hugoderwolf]

So I am in the middle of testing a small synth I have developed, and I'm referencing my output gain with other synths.

Please don't do that, most synths these days are way too loud. Massive is really evil in this regard, as is Logic's Retro Synth. I mostly have to turn those down by >30dB to get them to reasonable working levels.

A good working level would be around -20dB RMS. You can't always control that reliably, but make it easy for the user to get to this area. You'll lose the loudness war, but people won't hate mixing so much.

[cixelsyd]

Maybe I didn't communicate myself accurately. I am calibrating my synth to a VU Meter. This means at peak volume using a single sine I am at 0db. I was using the other synths as a reference for how the output fluctuates when multiple notes are played.

I agree that louder is not better. I much rather have awesome clarity. I also want my synth to be "sociable" on its intended platform.

Do you guys think I should be calibrating my synth to something other than 0DB?

[fluffy_little_something]

In terms of recording it would actually be good to have the same output volume no matter how many notes you play. If you want the chord to be louder, you can always hit the keys harder

Or how about letting the user decide, i.e. implementing a control or modulation for that: at zero the output is the same regardless of the number of notes played, and at max the volume adds up with every additional key.

[cixelsyd]

I like the idea of letting the user decide. If I'm not mistaken your idea for implementation has a similar effect of a compressor. with this in mind I can see BertKoors logic now which is also "let the user decide".

[BertKoor]

In terms of recording it would actually be good to have the same output volume no matter how many notes you play.

That is actually a VERY bad idea, not at all how any instrument in the real world behaves. Adding a note never makes another one softer.

[fluffy_little_something]

In terms of recording it would actually be good to have the same output volume no matter how many notes you play.

That is actually a VERY bad idea, not at all how any instrument in the real world behaves. Adding a note never makes another one softer.

I know it is not natural, but that doesn't mean it is bad. It would make recording easier. One could still achieve certain volume changes via velocity.

I think low notes should be louder than high notes because one usually doesn't play chords on low octaves as it results in an ugly mess.

[BertKoor]

I think low notes should be louder than high notes because one usually doesn't play chords on low octaves as it results in an ugly mess.

In a modular synth you can get that with reverse of key tracking modulating the VCA/DCA.

[JCJR]

In terms of recording it would actually be good to have the same output volume no matter how many notes you play.

That is actually a VERY bad idea, not at all how any instrument in the real world behaves. Adding a note never makes another one softer.

I don't disagree with your fundamental points, but there are some real world insts that don't increase +3 dB per doubling of voices.

The "passive mixing" nature of hammond tonewheel organs, because generators are cross connected to multiple keys, a typical big chord doesn't scale in that fashion compared to single notes or small chords.

Air powered organs, reed accordions, melodica, harmonica, might also get "diluted" in volume with added notes, if the air supply stays about the same but the number of notes is increased.

[AdmiralQuality]

Synth voices are typically mixed just like mixer channels, additively. If there's some limiting or compression going on then that's probably an effect setting in the patches of the synth you're testing against. A synth playing 2 voices shouldn't be any more or less loud than two separate instances of the synth each playing single voices, and mixed together at 0 dB gain. (Though, yes, phase correlation or lack thereof can make all the difference here.)