Skip to main content

Topic: Flac, replay gain and ...... (Read 5378 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
Flac, replay gain and ......
Hey , can anyone help me? i use foobar ,and  have tons of lossless files, i see the function replay gain but im just learning to use it , heres what ive done to tracks that sound way too low , ive increase the decibels and converted the flac file to flac (click on file > convert to> replay gain > move the preamp together> convert), so when i do that is the process irreversible?
and can someone teach me how to apply replay gain? and explain this thing to me ?
Note: since im a noob , what i do is keep my replay gains between -6db to -9db , there are some track like "one direction - whats makes you beautiful" that are about -12 db(that too darn loud), and the "killer - mr brightside" is that loud too but its not bad because it is rock (rock known for distortion) but the OD song just sound like a hiss fest(the person who produce it should go back to mastering or producing school), also the track i mentioned ive turned them down using the method above but i wanna know if that "technically" degrades the song
 

  • dhromed
  • [*][*][*][*][*]
Flac, replay gain and ......
Reply #1
If you've added some replaygain to the file, there's no need to convert anything. The value is a tag and foobar2k will use that to correct the volume on the fly.

The question is: do you want to permanently and semi-irreversibly apply the gain to your files? There's probably no point to this, unless you also use another player, like a portable one, that doesn't support replaygain tags.

  • yourlord
  • [*][*][*][*]
Flac, replay gain and ......
Reply #2
The idea behind replaygain is you decide on a reference level you want your library to conform to. When you apply replaygain to a file your software should analyze the average levels of the signal and then calculate a gain level to be applied to make that track (or album) average to the desired level. It then adds one or more tags to the file that simply defines the amount to adjust the level of the output so that the average output level is what you want.

It does NOT change the stored audio. You can revert the file to normal by simply removing the replaygain tags. If you play the file on a player that doesn't support replaygain it will play at the original mastered level.
  • Last Edit: 01 November, 2012, 11:17:39 AM by yourlord

  • db1989
  • [*][*][*][*][*]
  • Global Moderator
Flac, replay gain and ......
Reply #3
To repeat what yourlord said…
The idea behind replaygain is you decide on a reference level you want your library to conform to. When you apply replaygain to a file your software should analyze the average levels of the signal and then calculate a gain level to be applied to make that track (or album) average to the desired level. It then adds one or more tags to the file that simply defines the amount to adjust the level of the output so that the average output level is what you want.
…and then to rephrase it: precisely the point of RG is to do the work of levelling them for you, and it does this scientifically rather than estimating as you have been…
heres what ive done to tracks that sound way too low , ive increase the decibels and converted the flac file to flac (click on file > convert to> replay gain > move the preamp together> convert), […] what i do is keep my replay gains between -6db to -9db
…so feel free to save yourself the work and let RG take care of it by using it as normal, which will be just fine unless you happen to differ from most other humans in how you perceive average loudness.

  • mjb2006
  • [*][*][*][*][*]
Flac, replay gain and ......
Reply #4
It may help the OP to realized that "apply ReplayGain" means something different than "add ReplayGain tags".

Normally you want to just add the tags. They tell a ReplayGain-aware player how much to adjust the volume in order to get to a reference level. The player will "apply" the volume change (if so configured) to the audio data on its way to your speakers or headphones. It won't be changing the audio in the original file.

If you "apply" the ReplayGain to the file, though, then this does change the audio data in the file, such that it will play back at the reference level no matter what player you use.

  • dhromed
  • [*][*][*][*][*]
Flac, replay gain and ......
Reply #5
precisely the point of RG is to do the work of levelling them for you, and it does this scientifically rather than estimating as you have been


I never agree with RG. While RG accurately tells me that some hypercompressed pop item needs a good -8dB smacking (I'm looking at you, Gnarls Barkley!  ), it also told me that another perfectly mastered bit of Motorpsycho needed -9, and I keep that on 0. It doesn't sound loud to me or hurt my ears at all. So that's why I personally don't scan RG except as an indication.

I imagine hardware is a big part of it: if my headphones somewhat attenuate the 3-8KHz range, then obviously most full-spectrum music like metal and rock sounds far quieter and smoother. Personally, I'm more sensitive to sharp attacks and decays and overemphasized vocals (and we're back to hypercompressed pop! And Tegan & Sara!), even if the music on the whole isn't that loud.

YMMV.
  • Last Edit: 02 November, 2012, 05:31:54 AM by dhromed

  • halb27
  • [*][*][*][*][*]
Flac, replay gain and ......
Reply #6
To me RG values are a good start for finding appropriate values. I change them for quite a lot of tracks. And yes, it also depends on the hardware. On my pc I listen with headphones, and optimizing for them doesn't necessarily mean things are also real good on my hifi system. But at that stage I rarely go back and change RG value.
  • Last Edit: 02 November, 2012, 08:32:08 AM by halb27
lame3995n -Q0.5

  • db1989
  • [*][*][*][*][*]
  • Global Moderator
Flac, replay gain and ......
Reply #7
dhromed: Interesting, thanks for the info on your experiences. Just in case, I didn’t mean to question anyone’s species by saying “most other humans”.  Out of interest, do your perceptions apply to the original algorithm, R128, or both?

  • dhromed
  • [*][*][*][*][*]
Flac, replay gain and ......
Reply #8
There's not much change between the two as far as I can tell. Maybe a dB or two. My experience often disagrees with RG by as much as 10.

Looking up the numbers right now, the Gnarls Barkley album The Odd Couple clocks in with corrections of -7 to a whopping -11 according to RG, but I've adjusted it to -3 across the board.

The relevant Motorpsycho album Trust Us has values of -8 to -10, and one +8 (!). I've adjusted none of it.
Another Motorpsycho album, Let Them Eat Cake, I've always perceived as rather loud, but here RG gives me numbers of -3, -5, -7 and the occasional -8. I've adjusted that to a flat -3 as well.

You can understand why my faith in replaygain is pretty much non-existant.  I do wonder, however, if my perception of loudness could be accurately predicted by an algorithm. Such a beast might have to combine the equal loudness curve with the dynamic range and try to measure attacks, if that's even possible.

Alternatively, I may just happen to be hypersensitive to the vocal frequency range. Sometimes a loud recorded singing voice cuts into my ears as ruthlessly as a pure sine wave. Is that easily testable?

  • smok3
  • [*][*][*][*][*]
  • Moderator
Flac, replay gain and ......
Reply #9
Players could have another loudness stage when user pin-points/presets each frequency and measure/apply all frequency points as equaly loud (this takes human + equipment nicely into one check), based on that replaygain/r128 can do its work. This may weirdly fail in multi-user listening situations.

p.s. Another thing that is not really calculated is level of love for a certain song (i guess "better" songs could need a dB or two).

So three stages:
1. measure/apply perceived loudness over the frequency spectrum
2. rg
3. love level
  • Last Edit: 03 November, 2012, 06:39:55 AM by smok3
PANIC: CPU 1: Cache Error (unrecoverable - dcache data) Eframe = 0x90000000208cf3b8
NOTICE - cpu 0 didn't dump TLB, may be hung

  • db1989
  • [*][*][*][*][*]
  • Global Moderator
Flac, replay gain and ......
Reply #10
dhromed: Yeah, I imagine the combination of differing responses in hardware and in the user could quite easily confound RG and other algorithms that assume uniform frequency response. This is like having to assume a normal distribution: too generous to the real world, which isn’t always willing to bend itself to suit a scientific approach.

Players could have another loudness stage when user pin-points/presets each frequency and measure/apply all frequency points as equaly loud (this takes human + equipment nicely into one check), based on that replaygain/r128 can do its work.
This is interesting! I wouldn’t be surprised to see something like this emerging eventually (somewhere).

Quote
This may weirdly fail in multi-user listening situations.
Not so weirdly: we should almost hope it would.

Quote
p.s. Another thing that is not really calculated is level of love for a certain song (i guess "better" songs could need a dB or two).
Haha… but I think this might take us on a circular trip right back to the beginning: the argument that that’s what volume controls are for.

  • dhromed
  • [*][*][*][*][*]
Flac, replay gain and ......
Reply #11
This is interesting! I wouldn’t be surprised to see something like this emerging eventually (somewhere).


A personal RG calibrator sounds like a perfect idea for a new foobar2K component.

Quote
Yeah, I imagine the combination of differing responses in hardware and in the user could quite easily confound RG


Even coming home from a loud day versus later at night, when all is quiet.

  • 2Bdecided
  • [*][*][*][*][*]
  • Developer
Flac, replay gain and ......
Reply #12
I do wonder, however, if my perception of loudness could be accurately predicted by an algorithm. Such a beast might have to combine the equal loudness curve with the dynamic range and try to measure attacks, if that's even possible.
The original ReplayGain algorithm includes an Equal Loudness curve, EBU R128/ITU 1770-2 just includes a bass shelf. The difference is significant with test tones, but far less so with most music.

p.s. Another thing that is not really calculated is level of love for a certain song (i guess "better" songs could need a dB or two).
I found that to be a 6 to 9dB effect!

Cheers,
David.

  • dhromed
  • [*][*][*][*][*]
Flac, replay gain and ......
Reply #13
For kicks, I noticed late last night that RG recommended me a -12dB (holy crap) for some Trail Of Dead songs even though I find them perfectly agreeable and don't modify their gain at all.

It looks like there is some correlation between how dense the music is and my non-perception of its loudness. I guess broadening the spectrum decreases the subjective loudness? It seems a little counterintuitive, but that's what I hear. As smok3 said, my love for these songs probably contributed to the lower subjective loudness. I hope this will not contribute to early deafness

  • probedb
  • [*][*][*][*][*]
Flac, replay gain and ......
Reply #14
There's definitely a market for someone to invent some sort of personal calibrator for sound  Mostly RG is fine for me, there's only occasional tracks/albums that seem to have been made too quiet so they play noticeably louder.

Flac, replay gain and ......
Reply #15
Thanks for all your help guys, it tried the update tags and remove tag and saw how its adjust the replay gian , but it seems the way i was doing it before wasnt bad