Audio level compression

tobyW

Member
Hi DanR, we've been round this one before, so I am disappointed to find that VRD6's audio level compression behaves like VRD5's.

In VRD5, I found that audio level compression probably suited recordings of meetings, but I couldn't get it to work for movies.

I attach a screenshot comparing VRD6 compression against VRD4, for a movie clip. Spot the VRD6 failings, no prizes!

Since most of my recordings are movies rather than meetings, I long ago gave up the struggle with VRD5 and reverted to VRD4.

Please tell me VRD6 has an option to use the algorithm from VRD4 ???

Cheers,
tobyW
 

Attachments

Danr

Administrator
Staff member
That's true, Version 6 didn't change the audio normalization compared with Version 5. As I may have mentioned before, the Version 4 algorithm could only handle stereo signals, and needed to be upgraded to support 5.1 channels. One of big differences between the V4 and V4/V6 is that V4 averaged over a 5-15 second rolling & smoothing window while V5/V6 averages audio over a 0.4 to 3.6 second rolling/smoothing window. This results in faster attacks and reaction times which is great for voice, but less so for movies and music. Did you try regular normalization rather than the adaptive one?

Once V6 has stabilized, especially the HEVC support, we can re-visit tweaking these settings.
 

Danr

Administrator
Staff member
TobyW, I looked at the code and it wouldn't be too hard to get the load the smooth parameters from a text file rather than being hard coded. If you're interested in playing around with them, I can send you a version that would let you customize them. We don't have the time right now to do the testing this requires, but if you're interested in doing the testing we can add this feature.
 

tobyW

Member
Thanks Dan. Yes that would be great, and I would be more than happy to test and explore it.
Regards, TobyW
 

Danr

Administrator
Staff member
No update. This one dropped off our radar, will have to put it back on after our next V6 release (not beta release).
 

tobyW

Member
That's a pity Dan. As it stands, it's a bit of a show-stopper.

"Dynamic levelling" with "Average level" set to -12db or quieter introduces serious overshoots, while "Normalise peak" is defeated by spikes in the audio file. A sorry pair of features indeed.

Levelling is also upset by a lengthy inaudible lead-in, but that scenario is probably rare enough to ignore, at least for now.

Anyway all is not lost! This is an opportunity to create the missing option "Compress and normalise"! It's as simple as setting the "Average level" to -6db, running the "Dynamic levelling", and then feeding the intermediate file to "Normalise peak".

The leveller gets rid of the spikes, and a -6db setting seems to quell the overshoots.

(Quote from Dan203 a long time ago: To deal with spike use the compression option. That will get rid of the spikes. Then you can normalize to get the average level where you want it. I'd recommend using PCM output for the intermediate step so that you can retain maximum quality before finally compressing to AAC.)

For a beta version, perhaps the user can do the 2 individual operations and feed the intermediate file to the normaliser. Then the only beta software change is to add the new option to the drop down list and set its level to -6db.

I'm tired and there's smoke in my eyes. Hope I got this right. Hey hydra3333, what do you think?

P.S. This is still not as good as VRD4, I'm just suggesting it as a temporary solution that would be better than nothing.
 
Last edited:
Top Bottom