Re: [Audacity-devel] Dynamic Compressor Effect

[Max M. <mm...@po...>]

Hi James,

it's no problem for me to wait for the 3.0.1 release.

The latency is mainly caused by various kinds of lookahead. One
lookahead is selected directly by the user, another lookahead is
implicit in case of the exponential fit envelope detector because the
algorithm needs to process the signal backwards in the attack stage.

Latencies in the range of minutes are extreme cases if users select
high lookahead times or, in case of exponential fit, high attack times.
But I don't want to limit the ranges of the lookahead and attack time
sliders for offline processing just because of latency in realtime
mode.

When using the analog simulation envelope detector together with low
lookahead (a few milliseconds) like a real analog compressor, there is
almost no noticably latency.

Max

On Saturday, 27 March 2021 at 14:41, James Crook wrote:
> Hi Max.
> 
> The previous update I had on this was about the long latency.
> Disabling realtime preview is sort of OK, but something seems wrong.
> A one minute latency is colossal.  That points to something
> underlying wrong.
> 
> Also 3.0.1 has now become very very much a maintenance release for
> addressing the 3.0.0 unitary project issues.  Paul would love to get
> some structural 
> changes in that improve independence between pieces of code, and I am
> saying no.
> We'd also like to get portaudio and FFmpeg library updates in, and I
> am saying no.
> 
> It is pretty clear that the release after 3.0.1 is going to be quite
> soon, so as RM 
> I'm sorry to have to say that your compressor will have to sit out
> for 3.0.1.
> 
> 3.0.1 is due to actually release on 17th April, so it's a further
> delay of about 3 
> weeks for you in seeing progress on your compressor getting in to
> Audacity.
> 
> Is there a good reason for the very long latency?
> 
> --James.
> 
> 
> 
> 
> On Sat, 27 Mar 2021 at 07:56, Max Maisel <mm...@po...> wrote:
> > Hi all,
> > 
> > I've polished my Dynamic Compressor effect and think it is ready
> > now
> > for inclusion.
> > 
> > Since my last mail (see
> > https://sourceforge.net/p/audacity/mailman/message/37109016/ for
> > the
> > initial message), I mainly reworked the user interface accoring to
> > your
> > feedback and fixed several bugs.
> > 
> > The effect is basically realtime capable but for now realtime
> > preview
> > is disabled after some discussion with Steve because there are
> > bad user experience due to high processing latency in the effect at
> > some settings.
> > 
> > Main probleme here is, that the RealtimeEffectManager gives the
> > effect
> > a small block, e.g. 256 samples, and expects the same amount of
> > samples
> > in return. So if an effect like my compressor has a high processing
> > latency, there can be an initial silence for over one minute which
> > the
> > users will think is a bug. Any changes made to the effect settings
> > will
> > only take effect after this long time as well.
> > 
> > A possible solution would be latency compensation in the
> > RealtimeEffectManager so that the effect can request the required
> > amount of samples in advance at the beginning or when the user
> > changes
> > settings. The silence at the beginning can then be discarded so
> > that
> > users do not notice it. But this is out of scope of the compressor
> > effect changes.
> > 
> > The lastest revision including CI builds can be found in my pull
> > request at https://github.com/audacity/audacity/pull/676.
> > 
> > The latest documentation for the effect is can be found at
> > https://alphamanual.audacityteam.org/man/Dynamic_Compressor.
> > 
> > I'm looking forward to further feedback and integration.
> > 
> > Max
> > 
> > 
> > 
> > 
> > _______________________________________________
> > audacity-devel mailing list
> > aud...@li...
> > https://lists.sourceforge.net/lists/listinfo/audacity-devel
> _______________________________________________
> audacity-devel mailing list
> aud...@li...
> https://lists.sourceforge.net/lists/listinfo/audacity-devel