Volume levelling and adaptive volume work differently when you play an album than when you play a mixed playlist. When you play a mixed playlist, volume levelling tries to get each track to approximately the same volume (called track leveling); if you used track levelling with an album, it would destroy inter-track dynamics (i.e. "soft songs" that the artist inserted to break up the flow of the album would sound just as loud as the loud songs). That's why volume levelling picks a single level for when you play an album and plays all the tracks at that level (called album leveling).
On the other hand if you enable adaptive volume in peak level normalize mode alongside volume levelling, JRiver looks at the whole playlist and normalizes the volume so that the whole playlist will be about the same volume, but also so that the loudest track in the whole playlist won't clip. If you have some very dynamic tracks in the playlist that will push down the overall volume.
So imagine you have an album with volume leveling targets of -8, -9, and -10. Also imagine you have a playlsit with those three tracks and one other track that only has a volume levelling target of -1. Assume all tracks have true peaks at exactly 0dBFS.
Here are the results in different settings:
1) Just volume levelling:
a) If you play the album all three tracks are -9
b) If you play the album in the mixed playlist, the first track will be -8, the second track -9, etc.
2) Volume levelling and adaptive volume
a) If you just play the album, peak level normalize will add back 8dB because that's the most that can be added without causing track 1 to clip, so your net adjustment is -1dB
b) If you play the playlist, peak level normalize will only add back in 1dB because the fourth track would clip if you added more than 1dB to it, so each track in the playlist would have a different net reduction (-7 for the first album track, -8 for the second album track, etc.)
Your mileage may vary by a dB or so, because as I recall adaptive volume leaves a 1dB "pad" for any potential intersample clipping and not all real-world tracks have true peaks at exactly 0dBFS, but you get the idea.
Author
Topic: Questions on Volume Leveling and Adaptive Volume output levels from Smartlists (Read 2757 times)