Normalising logic

[drystream]

I routinely normalis[z]e my recordings to -1dB, but have wondered just how the process is designed.  Sometimes an album will start off really strong and I might get a couple of clips so I'll move the slider down a notch.  However, then several following songs might be lower in level, to the point where it seems like the slider should be moved back up.  Albums can be quite variable, so it's not practical to monitor the whole thing before recording to ascertain the peak level.  I'm also sometimes away from the computer while recording sides, so variations in volume aren't always noted.

Does VS read in the whole file for each album to determine the adjustment to -1dB?  Or are the track breaks taken into account to normalize each track separately?  The real question here is how to best normalize to end up with all the tracks being approximately the same volume with respect to other tracks from other albums.  I realize, of course, that some tracks are intended to be quieter than others on a given album, so you don't necessarily want each track to blast out at the same level.  I should add too that I use Pure Music for playback, so the volume adjustment within iTunes can't be used.

Of most concern is in the situation where two or even three recording levels are used during the recording of one album--does this compromise the normalizing process?

[Paul Sanders (AlpineSoft)]

Hi,

VinylStudio normally reads the entire album side to determine the peak level.  If you subsequently find that a particular track is too quiet, you can select it and normalise it individually.  You do this as follows:

1.  Add a new filter set covering the tracks in question

2.  Click 'Normalise'.

You add a new filter set via the 'Add Filter' button under Advanced Options.

There's no way to tell VinylStudio to normalise on a per-track basis automatically.  Perhaps there should be.

[drystream]

Thanks Paul--that's just what I've been doing to fix tracks that are either too loud or too quiet compared to the rest of the album.  I was just curious how VS works in evaluating entire albums.  I'm still very happy with the versatility and flexibility of VS in handling problems.   I'm up to about 650 albums recorded/processed so far.

[Paul Sanders (AlpineSoft)]

OK, well it normalises relative to the peak level of the entire album side, although there is some logic to ignore transients. How effective that logic is I'm not sure...

[drystream]

Paul, not to be too obsessive about this, but what does the 'gain' number mean?  I assume that it is in reference to some baseline, so how is the baseline established or defined?  I almost always end up with a gain number between about 1.2 and 1.7 or 1.8, no matter how 'hot' some of the tracks are, even with a handful of clips, when normalizing to -1dB.  If I go to -2dB or -3dB then the gain number is < 1.0.

[Paul Sanders (AlpineSoft)]

The gain is the amount by which the digitised signal will be boosted (or cut) as a result of normalisation.  Thus if you had of a gain of 1.2, a sample with value of 0.5 (on a scale of -1 to +1) would become 0.6, and so on.

[sturbee]

Hello All, What I do is rip the LP to FLAC using VinylStudio, next I use dbPoweramp Batch Convert to normalize the FLAC files, this will flag each file with a db plus or minus and have set it to also normalize the album. One must add DSP Effects/Actions of "Replay Gain" (do not use "Replay Gain Apply") to the dbPoweramp Batch Convert program. I also use Mp3tag to check art work and touch up tags. I used dbPoweramp to rip over 800 CD's to FLAC then stored these files on a vortexbox server for use with a Logitech Squeezbox Touch. Ripping the same CD as the vinyl and comparing the two for volume, I did not detect any difference in volume.