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

[Calagan]

Hi There everybody

Recently, while testing some 3d party plugins, I discovered that Ableton Live 11.1 was gating the audio output of 3d party plugins below a certain level.

Under -110db RMS, the output of any 3d part plugin (VST2 and VST3) is gated and disappear, like if the dynamic range was reduced to 110dB.
No issue in Reaper, and everything is ok after built-in Live devices...

I did a video (poor quality and no audio, sorry) to demonstrate it. Please feel free to check it and tell me what you think about it.
https://youtu.be/RpU_5dXa1es

I also sent an email to the Ableton Live support.

Do you have any explanation about that ?

[[jur]]

Are you seeing different results in eg. L11.0?
Which bit depth setting are you using?

[Calagan]

Are you seeing different results in eg. L11.0?
Which bit depth setting are you using?

No relation to bit depth setting : it's only for recorded audio.
And I don't want to install previous versions to test. I'm waiting for some feedback from users..

[[jur]]

Honestly I don't know why this is happening, or maybe this has to do with the feature that bypasses plugins processing when they receive "silent input" to save CPU.
but I'm wondering why does it matter to you? In which situation do you need a -146dbFS signal?

[miyaru]

Honestly I don't know why this is happening, or maybe this has to do with the feature that bypasses plugins processing when they receive "silent input" to save CPU.
but I'm wondering why does it matter to you? In which situation do you need a -146dbFS signal?

Oh, that is not the point..... The question was does Ableton Live do this and if yes why. You never use soft signals like that, but it might be a bug, ain't it?

[Calagan]

Honestly I don't know why this is happening, or maybe this has to do with the feature that bypasses plugins processing when they receive "silent input" to save CPU.
but I'm wondering why does it matter to you? In which situation do you need a -146dbFS signal?

Oh, that is not the point..... The question was does Ableton Live do this and if yes why. You never use soft signals like that, but it might be a bug, ain't it?

You never use soft signals like this, but actually it's like processing the whole audio in more or less 20 bits (instead of 32 bit floating point).
I'm not very technical, but I guess there's a lot of very good reasons why 32 bits is used instead of 24 or 16 bit depth processing...

[[jur]]

That's why I asked about the bit depth resolution. I'm wondering what's the resolution of the plugins you're using in the video. Could they be processed at a lower resolution compared to Live native devices?

[Calagan]

That's why I asked about the bit depth resolution. I'm wondering what's the resolution of the plugins you're using in the video. Could they be processed at a lower resolution compared to Live native devices?

Pro Q3 is processing at 32 bit depth resolution.
Equick and TrackComp are using 64 bit depth resolution.
SPAN is supposed not to modificate the audio and pass it through...
Live is supposed to use 32 bit depth resolution at each point of the signal path, and 64 bit depth at summing points...

By the way, the exact same signal path (with these 3d party plugins) doesn't show any issue in Reaper...

So there's clearly a big issue... It's not even 24 bit depth dynamic range : -110db is between 16 and 24 (around 20 bit, I don't know how to calculate).

[[jur]]

Well, that's probably worth an email to support@ableton.com
On my end, I'll forward the info to the team

[bpm.qa]

Hi Calagan,

As [jur] mentions, please contact our Tech Support team for further investigation.

Thanks!

[Calagan]

Yes of course. I did contact the support at the very beginning (it was the first thing I did)
It's been 5 days and I didn't have any answer.

I understand it can take long time, so I don't complain but I was thinking that maybe someone here can understand what happens...

[Calagan]

By the way : can someone test their own copy of Live and confirm they have the same issue ?
I opened a thread on Gearspace, a guy answered and he did not find the same results : no gating for him.
It was not clear if he followed the same test protocol.
Very strange indeed...


EDIT : in order to test, it's very easy : create a continuous noise (white or pink) and find a way to go as low as -110db. Use at the output a monitoring plugin that can see such low level.
Use any 3d party plugin with no processing between the noise source and the monitoring plugin and see if it gates the audio below a certain level.

[just_in]

I can reproduce this in both 11.0 and 11.1 using some white noise, Utility and SPAN:

With VST off:


With VST on:


The screenshots should be self-explanatory and easy to reproduce, but I can post the white noise file if needed.

[fishmonkey]

there is definitely something funky going on. and if you go down to about -126dB the signal cuts in and out (tested with pink noise and Voxengo Span Plus on Live 11.1).

[Calagan]

Thanks all for the confirmation of the issue.

It's strange because in my tests, the threshold is around -106db RMS with pink noise. Almost 20db louder than the fishmonkey test.
Maybe there is some kind of frequency dependant gating, or maybe the issue is related to something else...

For anyone willing to test, I uploaded a very simple test Live set in a thread I opened on Gearspace (the only requirement is using the free SPAN plugin) :
https://gearspace.com/board/ableton-liv ... st15886514