Live 11.1 is gating low level audio after 3d party plugins

[Calagan]

By the way, is -106db meaning anything to you audio geeks ?

I know 144db is the dynamic range of a 24 bit depth processing, and 96dB is the dynamic range of 16 bit depth processing.

If we had these numbers, I could relate the issue to something I know. But 106 db ???

[fishmonkey]

i just had another look at this and with the exact same setup i'm no longer seeing the issue. very strange.

ah! it wasn't the exact same test. the problem doesn't appear using the Audio Units version of Span Plus. it does appear when using the VST and VST3 versions.

[Calagan]

i just had another look at this and with the exact same setup i'm no longer seeing the issue. very strange.

ah! it wasn't the exact same test. the problem doesn't appear using the Audio Units version of Span Plus. it does appear when using the VST and VST3 versions.

Interesting. I never use AU. So it seems there is an issue with VST and VST3...

[Calagan]

Actually, it looks like it's an old issue that was never fixed : https://gearspace.com/board/music-compu ... gnals.html

[just_in]

Interesting.
Well, that Options.txt workaround doesn't work anymore. But this should just be fixed already.

[[erm]]

This is currently being looked into.

[Calagan]

This is currently being looked into.

What do you mean, "looked into" ?
Are you speaking about the Ableton team checking the issue ?
How do you know ?

[[erm]]

Yes Ableton is investigating the issue, and I know because I work for Ableton.

[Calagan]

Yes Ableton is investigating the issue, and I know because I work for Ableton.

This is great news ! I hope you'll find an easy fix.

Do you know what audio consequences it can have in a real life mixing scenario ?
Of course I understand it affects reverb tails and fades, but maybe there are more consequences I'm not conscious about...

[Calagan]

Ok, I just received an answer from the Ableton support :
"This is known and expected behavior, there is a built-in gate function that cuts off the audio of plugins below -100dB to save processing power. In certain scenarios, this can have unwanted (not )audible effects, e.g. when intentionally amplifying/compressing the signal afterward."

They sent me the option.txt I can use to change this behavior, but it seems it's intended to work like that.
I find it very strange because actually this is like reducing the dynamic range.

[[jur]]

I find it very strange because actually this is like reducing the dynamic range.

Who cares, it's all about the loudness war anyway
But more seriously, nobody can hear such low levels so...

[just_in]

Ok, I just received an answer from the Ableton support :
"This is known and expected behavior, there is a built-in gate function that cuts off the audio of plugins below -100dB to save processing power. In certain scenarios, this can have unwanted (not )audible effects, e.g. when intentionally amplifying/compressing the signal afterward."

They sent me the option.txt I can use to change this behavior, but it seems it's intended to work like that.
I find it very strange because actually this is like reducing the dynamic range.

Can you post the option.txt setting(s)? I tried that one mentioned on Gearspace and it didn't work, but maybe I made a mistake.

Frankly, I don't think this is OK. They can put some "CPU saving" switch in Live for anyone who wants to use it, but by default Live should process audio with full precision.
And it makes no sense that they would only disable processing for VSTs, but not for other kinds of plugins (AU and native).

[Calagan]

Yes, I find it very strange and, actually, shocking !

It's like reducing the dynamic range. I don't know if it has audible effect, but it's not how processing audio in 32 bits is supposed to behave.
I know it's not like truncation, but I would prefer to choose what happens instead of not knowing it.
For exemple you could have an option in the preference panel, and choose "save CPU" while mixing and "full dynamic range" while exporting...


Regarding the option txt, file, I can't figure out how to attach a file, so what you need to do is to create a txt file called "Options.txt" and fill it with the following text :

-_AudioOffTimeInSamples=691200000

[just_in]

Agreed, not much else to add.

But I'll just add this again: that Options.txt setting does not work. And I do have the Options.txt file set up correctly, because I can successfully change some other settings with it, but -_AudioOffTimeInSamples=691200000 does nothing.
In fact, any line that starts with -_ is completely ignored by Live, no errors pop up even if I write something random.
I'm on Windows, by the way. Maybe it works on Mac...

[[jur]]

I'd be curious to see (hear!) a 100db dynamic range audio file if you have one to share.
32 bits floating point is mostly relevant for the highest part of the dynamic range (goodbye clipping!). It can also be useful for the lower side of the range but "only" when recording very high dynamic range stuff, so that you can then boost the quiet parts without boosting the noise (well, if your recording equipment can do it)... but you'll boost those way higher than -100db anyway.
I get it that theoretically it's like destructive compression (compression in the sense of data compression/truncation) but I really really doubt that it can have any bad side effect in practice (except for the scenario mentioned by support, which is solvable...).

I didn't test myself but I think someone said that it doesn't happen with native Live devices? If that's right, then that's one another good reason for sticking to them as much as possible