View Single Post
Old Aug 10, 2015, 06:07 AM
LiquidAcid LiquidAcid is offline
Trusted Editor
Join Date: May 2008
Posts: 1,574

Just did a little analysis I wanted to share here.

I computed the dynamic ranges (according to EBU R128) of the tracks:
101 Pregap: replaygain_track_range=0.00 LU
102 Serenity: replaygain_track_range=22.31 LU
103 Wailers And Waterwheels: replaygain_track_range=19.21 LU
104 To the Sun: replaygain_track_range=17.02 LU
105 A New Hope: replaygain_track_range=19.13 LU
106 On Westerly Winds: replaygain_track_range=17.25 LU
107 I Am the Sea: replaygain_track_range=15.28 LU
201 Pregap: replaygain_track_range=0.00 LU
202 Primal Judgment: replaygain_track_range=1.51 LU
203 Under the Weight: replaygain_track_range=1.06 LU
204 Fallen Angel: replaygain_track_range=3.29 LU
205 Through the Maelstrom: replaygain_track_range=1.87 LU
206 Thunder Rolls: replaygain_track_range=3.79 LU
207 Oblivion: replaygain_track_range=0.76 LU
Disregard the "Pregap" tracks, they're just an artifact of the initial extraction from the BluRay.

The audio has a sampling rate of 96kHz and a bitdepth of 24. For comparison, Redbook content on a regular compact disc has a sampling rate of 44.1kHz and bitdepth of 16. In particular Redbook allows for a wooping dynamic range of 90dB. If you apply some safety tresholds, then of course you get less then 90dB, but just keep that 90dB in mind for now.

With even more bitdepth, the question arises if the mastering engineer made use of it at all? Well, the answer is no. While for the piano arrangements we see a maximum range of around 22dB, which is not great but OK, the rock/metal remixes go as low as 0.76dB. That's not even scratching our 90dB from above.

Maybe the increased bitdepth wasn't used, but at least the increased sampling rate was good for something? I computed spectograms with SoX using default settings with doubled resolution (sox file.flac -n spectrogram -x 1600 -Y 1100 -o spectrum.png).

You can find the images here.

The piano arrangements look particularly good, even though they of course don't reach 48kHz at all. The rock/metal tracks on the other hand show the typical cutoff at 24kHz. The energy above this treshold is probably moslty due to 'reflection' artifacts (see artifacts of linear resamplers here e.g.).

So even on the sampling rate side the engineer didn't make use of the increases range.
Reply With Quote