ReaXcomp misbehaves with automated threshold...

[Fabian]

It seems that ReaXcomp misbehaves when having its threshold automated.

In the Licecap below, I automate band 2 to compress pink noise in the Comp region, and we can see that at the transient where the envelope goes to -14, ReaXcomp briefly compresses the signal -11 dB. If I just let ReaXcomp compress band 2 of the noise, not playing, it compresses between -2.5 and -4.7, nowhere near -11. And this misbehavior is audible, and can also be seen in the rendered output.



ReaXcomp settings:
Attack 0ms
Release 100 ms
RMS 5 ms

[Fabian]

No one else but me?
Maybe it is just here on my install, then...

If I fix the threshold and instead automate the ratio, everything seems to work fine. But nor when automating the threshold, as we see above.

And I forgot to mention:
Reaper v7.14 portable, W10

[SMM]

Originally Posted by Fabian

It seems that ReaXcomp misbehaves when having its threshold automated.

Speaking of ReaXcomp....

I have always wondered why there is bit activity (stuck bits?)
in ReaXcomp when no audio is being processed in the plugin.

I attached a screen shot.

Do you think that this might be effecting things
when you are trying to automate your settings?

I'm guessing that the result in my attached image
has no effect on your results when you are automating your settings.

However, I'm posting this here in the event that there is a chance that
it could be related in some way?

[Fabian]

Originally Posted by SMM

Speaking of ReaXcomp....

I have always wondered why there is bit activity (stuck bits?)
in ReaXcomp when no audio is being processed in the plugin.

Yeah, that's another strange thing. Something comes out of ReaXcomp even when all bands are inactive and there's no input. I can confirm that.


(Click pic to view larger animated gif)

It's probably not related to the issue I reported, strange and unexpected nonetheless.

[SMM]

Originally Posted by Fabian

Something comes out of ReaXcomp even when all bands are inactive and there's no input. I can confirm that.

It happens when moving the thresholds up and down also
(when the bands are active).

I use ReaXcomp in my productions.

I render my mixes to a 64 bit float stereo wav file (master archive file)
before I truncate to 24 bits and 16 bits.

When I truncate to 24 bits and 16 bits,
those low level "stuck bits" are discarded.

I wonder if this is a programming bug?
Perhaps the both of us have discovered two issues
that need addressed and fixed in regards to ReaXcomp?

Note: I use REAPER version 4.75 and version 7.11 and this
issue is present in ReaXcomp for both versions.

[Justin]

Originally Posted by Fabian

Yeah, that's another strange thing. Something comes out of ReaXcomp even when all bands are inactive and there's no input. I can confirm that.


(Click pic to view larger animated gif)

It's probably not related to the issue I reported, strange and unexpected nonetheless.

it adds some anti-denormal noise in there

[Justin]

Originally Posted by Fabian

It seems that ReaXcomp misbehaves when having its threshold automated.

In the Licecap below, I automate band 2 to compress pink noise in the Comp region, and we can see that at the transient where the envelope goes to -14, ReaXcomp briefly compresses the signal -11 dB. If I just let ReaXcomp compress band 2 of the noise, not playing, it compresses between -2.5 and -4.7, nowhere near -11. And this misbehavior is audible, and can also be seen in the rendered output.



ReaXcomp settings:
Attack 0ms
Release 100 ms
RMS 5 ms

This quirk also exists when automating ReaComp's threshold. Hell, throw ReaComp on some silence and wiggle the threshold around and watch the GR meter kick in, lol.

Fixing these (contemplated some buried compatibility flag for existing projects, though seems like an obscure enough and undesirable behavior, so maybe not...)

[SMM]

Originally Posted by Justin

It adds some anti-denormal noise in there

Thank you for this information.
Much appreciated.

[SMM]

Originally Posted by Justin

This quirk also exists when automating ReaComp's threshold. Hell, throw ReaComp on some silence and wiggle the threshold around and watch the GR meter kick in, lol.

Just for confirmation purposes:
I do not see this happening in version 4.75 (gain reduction meter kick in).
However, I do see it happening in version 7.11 as you described.

[Daodan]

Originally Posted by SMM

Just for confirmation purposes:
I do not see this happening in version 4.75 (gain reduction meter kick in).
However, I do see it happening in version 7.11 as you described.

There is no gain reduction meter in v4

[SMM]

Originally Posted by Daodan

There is no gain reduction meter in v4

There is a gain reduction meter in v4.75.
(see the attached image)

[Daodan]

Originally Posted by SMM

There is a gain reduction meter in v4.75.
(see the attached image)

Stupid me. Thought you were talking about GR meters in TCP/MCP.

[Justin]

Originally Posted by SMM

Just for confirmation purposes:
I do not see this happening in version 4.75 (gain reduction meter kick in).
However, I do see it happening in version 7.11 as you described.

It happens in 4.75, it has happened in ReaComp since 2007 (though maybe the GR meter misses it). If you automate the threshold on a signal you'll see it. anyway should be fixed in the coming +dev builds.

[Fabian]

Originally Posted by Justin

It happens in 4.75, it has happened in ReaComp since 2007 (though maybe the GR meter misses it). If you automate the threshold on a signal you'll see it. anyway should be fixed in the coming +dev builds.

Great!
Thanks.

[SMM]

Originally Posted by Justin

It happens in 4.75, it has happened in ReaComp since 2007 (though maybe the GR meter misses it). If you automate the threshold on a signal you'll see it. anyway should be fixed in the coming +dev builds.

Thank you for the information.
Thank you for clarifying.
Thank you for the fix.
Thank you for all that you do with REAPER, you are the best.
Thank you.
Much appreciated.