Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Why is ReplayGain's Album gain tag necessary? (Read 2477 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Why is ReplayGain's Album gain tag necessary?

Hello, this is my first post but I have a question.

As I understand it, ReplayGain's album gain tag is calculated by concatenating the entire album into a single song, and then calculating the loudness gain.

But is the result of that method significantly different to just getting the largest loudness gain value of a track from that album, and applying that to all the tracks? After all, it seems logical that both methods compensate how the loudest part of that album should be normalised. So is there any point to separately concatenating, calculating and storing an album gain?

Thanks!

Re: Why is ReplayGain's Album gain tag necessary?

Reply #1
The way I think of it - replaygain "allows" a certain amount of "loudness" to each album.  If an album is brickwalled, then it will all play at exactly the average loudness.  If an album has some loud tracks and some quiet ones, then the loud tracks are allowed to be louder than the average (louder than the brickwalled album) and the reverse with the quiet ones.

The way you describe it - you're interpreting it as a limiter. It doesn't work that way. It equalises average album volume, not peak album volume.

I think albumgain is great - I never use trackgain. Because if a track is supposed to be quiet in the context of an album, then I imagine it should be quiet in the context of whatever playlist I put it in.

Re: Why is ReplayGain's Album gain tag necessary?

Reply #2
But is the result of that method significantly different to just getting the largest loudness gain value of a track from that album, and applying that to all the tracks?

The "gain" is not the "peak". The album "peak" works like you describe. The "gain" is akin to a weighted average loudness.
If you include one more track that does not have as loud a peak, album peak does not change, but album gain (typically) does.

So is there any point to separately concatenating, calculating and storing an album gain?

Yes. Unless you want volume to boost up at track transitions within the same album - say you listen to a live album, then what? Most of us want to keep album volume constant if we listen to an album. That is why we have the album tags. If you play just single tracks one from each album, then you probably want the track gains.

If you have a "various artists" sampler, then maybe you don't want that particular album to have album gain/peak. Then you don't need to scan as album - and if you already did, you can delete the album gain and album peak tags later.

Re: Why is ReplayGain's Album gain tag necessary?

Reply #3
And to give some actual numbers how much difference there can be between the loudest track gain vs album gain I found the highest difference from my FLACs to be 21.64 dB. Highest difference between lowest replaygain track gain value and album gain value in this set was 9.25 dB.

Re: Why is ReplayGain's Album gain tag necessary?

Reply #4
"The Misinterpretation of Silence and Its Disastrous Consequences" by Type O Negative could be a telling track title, uh?
On "The Least Worst Of" (the short version!) it peaks at 0.012 with a track gain of +34 (album gain: -7.74). There is just some low noise down there, likely not meant to be heard.

21.64 isn't quite unique. Fantômas' s/t album "kind of omits" track 13, it being a lucky number, and track 13 contains the fade-out of the final cymbal crash from track 12.
Album gain: -6.29. Track gains: -4.03 for track 12 and +26.97 for track 13. Using track gain would boost the cymbal by 31 dB during fade-out.

A difference of over 30 dB I can also find in a "stand-alone track" with actual music on John Zorn's "Kristallnacht" album. Album gain: -7.11. Track 2 ("Never again", with all the sound of breaking glass): nearly twelve minutes at -9.31. Then the contrast to track 3 "Gahelet (Embers)": three minutes and a half at a track gain of +23.45.

All FLAC, so no lossy-encoding artifacts here.

Re: Why is ReplayGain's Album gain tag necessary?

Reply #5
I might delete the album gain or track gain values from an album if either is out of line and shouldn't be used. For example, a segued gapless album that is to be listened in one piece might not contain a track gain in my collection, or if it is a compilation or a single with unbalanced loudness levels seemingly in error, I might delete the album gain, and have the player fall back to track gain automatically.

 

Re: Why is ReplayGain's Album gain tag necessary?

Reply #6
I see, as I understand it now the album gain is the average loudness of the album compared to the reference loudness.

One thing I didn't understand was why the average of track gains (average of average loudness) does not exactly equal the album gain. One reason is that ReplayGain performs statistical processing such that the loudest 5%-th energy level is used. So obviously the average of loudest 5%-th of all tracks will not be the same as the average loudest 5%-th of an album.

I was asking this question because some people have said that iTunes' Sound Check does Album Gain, but the iTunNORM tag only has room for a track gain. So I was wondering if maybe they derive the album gain from track gain, and whether that would yield comparable values to ReplayGain's album gain.

I can definitely tell that a track that has a Sound Check gain of -0.1dB as seen from iTunes increases in volume noticeably with Sound Check on when I play it inside an album (playing it from it's album track list). To do this, I play the music, go into settings, and repeatedly tap the Sound Check setting so I'm sure I'm hearing a difference, although I won't be able to objectively confirm until I get some 3.5mm to laptop mic cables.

I can also hear a comparable louder volume with Sound Check on compared to off in tracks with Sound Check gains +21.5dB and +1.2dB (track gain values from iTunes, same album, album playback), indicating that Sound Check track gain value doesn't seem to play a significant role in album playback.

I can also tell that there is no noticeable difference in volume when I play them from a playlist, between Sound Check on or off. So the iPhone must treat album playback differently and there must be some different gain value that it uses when inside an album playback mode, because I think what difference I hear in album gain and album playback is much greater than what I hear from the playlist and the 0.1dB track gain difference.

What's also interesting is that this album's average track gain value is +6.6dB (simple average of iTunes reported gain values), and when I try to play a track with a Sound Check gain value of +2.0dB with Sound Check on, I'm pretty sure I can hear a noticeable increase in volume in album playback (what I think would be +6.6dB) compared to playlist playback (what is +2.0dB), although again I'm not sure if there's any real correlation unless I can measure the difference.

Another I guess interesting tidbit is that unlike ReplayGain v1, the more recent ITU-R BS.1770 standard does NOT use statistical processing, and the gain value is indeed the average loudness over the entire track (SEE: https://www.dolby.com/uploadedFiles/Assets/US/Doc/Professional/AES128-Loudness-Normalization-Portable-Media-Players.pdf Section 3.2.2.1). In that case, I think averaging the track gain values with some weighting to obtain an album gain would be a perfectly legitimate technique?