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: Replaygain applied files not limited to same volume (Read 10913 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Replaygain applied files not limited to same volume

I just reformatted my PC and moved from 0.9.1 to 0.9.2. My media is all tagged FLAC with replaygain tags. Most things work but recently I've noticed that while watching the output/peaker meter on my soundcard (E-Mu, using ASIO drivers), about 95% of my music is limited to what looks like -10dB, the way I expect with replaygain applied. But with some albums the limit is most definitely a good 5dB louder.

Correct me if I'm wrong, but isn't the whole aim of using replaygain "album gain" to keep all the dynamics within the albums intact, but for the maximum volume of all albums to be scaled to the same absolute volume (I'm using the default 89dB), so that you don't have to go jumping for the volume control every second album, and so that newer "hot" mastered albums sound cleaner (at least that's what I've found from listening).

The fact that the volume scaling by RG is obviously (visually and by listening) applied perfectly to 95% of my music makes me think some strange is happening with the few albums that don't appear limited (and also that the rest of my audio setup, eg soundcard drivers etc are configured fine).

This really bugs me, as RG used to be great and I never had to touch the volume, now I have to tap it down a few dB on some albums. Very strange. I suspect that more information would be required to solve this problem but I'm not sure what's desired. Just reply with what you want to know, hopefully someone has an idea! The rest of my audio setup is exactly the same as previously (ie before the reformat and before upgrading Foobar). And I'm 100% sure the problem doesn't lie in another part of the setup, since as I said before nearly all the music is fine.. just the odd album isn't limited and sounds too loud).

I've rescanned files which had their RG info originally applied by 0.9.1 and they get the same values for album and track. I've also (just now) downgraded back to 0.9.1 and it makes no difference between versions. Again leads me to think something strange has happened with the RG tags. But one of the "bad" albums was created yesterday via:
EAC>WAV, Foobar>FLAC, Foobar> + RG tags.

Thanks for any help in advance.

Replaygain applied files not limited to same volume

Reply #1 –
are you using "track gain"... or "album gain" mode for playback...?

with track gain I would expect them to remain fairly consistent... but with album gain there could be differences...

Replaygain applied files not limited to same volume

Reply #2 –
Quote from: rosshmusic on
are you using "track gain"... or "album gain" mode for playback...?

with track gain I would expect them to remain fairly consistent... but with album gain there could be differences...


Album gain. But I disagree with you. The albums should have the same *peak* volume using album gain RG. If you turn off RG and just set foobar's volume to -10dB (on tracks with and without RG), that would get rid of the "hotness" from newer albums and clean them up a bit, but it would also make the "good" old albums much quieter at the same time. RG adjusts those albums that actually need the cut, and leaves (or only slightly adjusts) those old albums that had a good overall volume (e.g. Most Dream Theater CD's get like a 0.5 or less cut, whereas any new generic pop usually gets -10db, and this happens on all my music too, as we'd expect). And the intention is that they all end up having the same 'absolute' maximum.

Track gain should not be used on albums at all, only on single tracks where it is impossible to calculate how "loud" they should be compared to the other tracks on the album they came from.

My problem is, although the RG numbers look right, some of my music is simply 5dB louder even using RG (viewing my output meters AND just using my ears). I'm still experimenting, but I don't know what else I can do without delving into code, a job which I think is much better left to those in the know

Replaygain applied files not limited to same volume

Reply #3 –
Try to lower the preamp for all files WITH rg to -5dB....... does the problem still happen? If no, then you are - like me - one of those users for whom the 89dB-target is too high.
I am arrogant and I can afford it because I deliver.

Replaygain applied files not limited to same volume

Reply #4 –
Quote from: sl33py on
Quote from: rosshmusic on

are you using "track gain"... or "album gain" mode for playback...?

with track gain I would expect them to remain fairly consistent... but with album gain there could be differences...


Album gain. But I disagree with you. The albums should have the same *peak* volume using album gain RG. If you turn off RG and just set foobar's volume to -10dB (on tracks with and without RG), that would get rid of the "hotness" from newer albums and clean them up a bit, but it would also make the "good" old albums much quieter at the same time. RG adjusts those albums that actually need the cut, and leaves (or only slightly adjusts) those old albums that had a good overall volume (e.g. Most Dream Theater CD's get like a 0.5 or less cut, whereas any new generic pop usually gets -10db, and this happens on all my music too, as we'd expect). And the intention is that they all end up having the same 'absolute' maximum.

Track gain should not be used on albums at all, only on single tracks where it is impossible to calculate how "loud" they should be compared to the other tracks on the album they came from.

My problem is, although the RG numbers look right, some of my music is simply 5dB louder even using RG (viewing my output meters AND just using my ears). I'm still experimenting, but I don't know what else I can do without delving into code, a job which I think is much better left to those in the know

I use track gain at work where I seldomly have whole albums to listen to... at home I use album gain... both are applied, I just set foobar to use the one appropriate to my location...

but IIRC, using album mode would not put all the tracks at the same peak (it attempts to maintain the dynamics between each track gain level)... so when played each would still have a slightly different level as they can not all be at the target and still maintains their respective differences... playing in track mode would remove those dynamic differences but play all the tracks at a consistent volume... please correct me if I'm missing something..

but it seems as though I may not have understood the stark differences you getting... is it precisely 5db... or does it vary slightly..?

Replaygain applied files not limited to same volume

Reply #5 –
I have one album in WMA and it sounds louder than other in 0.9. This is the only music I have in WMA, so I thought this was something with WMA playback. After reading this thread, I took a look and it also sounds like 5dB difference.
Maciek

Replaygain applied files not limited to same volume

Reply #6 –
@ setting it to -5dB even for RG'ed files. This will not work because then the 95% of my albums that sound fine will now be a bit quieter than the 5% of albums I'm adjusting the volume for! Thanks for the suggestion anyway.

@ but IIRC, using album mode would not put all the tracks at the same peak (it attempts to maintain the dynamics between each track gain level)... so when played each would still have a slightly different level as they can not all be at the target and still maintains their respective differences... playing in track mode would remove those dynamic differences but play all the tracks at a consistent volume... please correct me if I'm missing something..

Sorry but I disagree with this completely. For about 95% of my albums, which have all kinds of dynamics in them, the maximum volume is most definitely limited to a point on my soundcard's output: 10 (I expect this means -10dB. And yes.. as you say when the music is quieter, the level is below this imposed limit. But this also happens with the other albums which (at their loudest) go over the limit. The dynamics are supposed to be kept as you say, but the maximum volume of all of the albums is meant to be the same in RG album mode, and it should be 89dB for all of them. Not 89dB for most, and different for some. If RG can get the absolute maximum volume perfect for 95% of files then I don't see why it shouldn't be able to get it right for 100%.

but it seems as though I may not have understood the stark differences you getting... is it precisely 5db... or does it vary slightly..?

OK I'll explain in more depth: Most of my music (albums) output a signal which, on my soundcard output, are visually (and just by listening) limited to a marker @ 10. it goes from 0 at the top to 10, then 20 etc. So it's -10dB I expect. And for most albums the signal NEVER ever goes above this line no matter how loud they were without RG.

But with these 5% of albums that aren't quite right: the volume hangs AROUND that same volume (-10dB) in general, but during stronger parts, it obviously peaks much higher than the 10 I see for everything else. Anywhere between 3dB and 7dB over. but usualyy the louder parts will br 5db louder than the limit imposed by RG on everything else. Sometimes it drops below the 10, but so does the "OK" music: this is just quieter sections of the music (dynamics). But in the "bad" albums, I don't believe the volume should go above the 10 marker, no matter how loud they are or what dynamics they have, as the other good albums range from metal to pop to folk to god knows what else, and they are all fine! Supertramp's Breakfast in America is sometimes too loud. Strapping Young Lad's self-titled album is fine. Most Kraftwerk is fine, but Tour De France Soundtracks goes over the limit. Metallica is fine, Infected Mushroom goes over (on several albums). From this we can see there is no real relationship between the type of music that is peaking too high and the stuff that is fine.

I've tagged and retagged, installed and uninstalled. Configured Foobar multiple times. Uninstalled my E-Mu drivers. It's really annoying as I want to see every file limited at that same ideal volume.

I thought it might have been the dreaded over compression of metal and heavier albums (eg Greenday or Machinehead) keeping their volume at that particular -10db level, but then Supertramp is a DCC Gold album and not compressed out the wazoo, neither is Infected Mushroom really. In fact, my Machinehead albums sound and look fine lol

Maybe some of u also have soundcards (better yet, the E-Mu 0404 like I have) where you can watch the playback levels of several of your albums. You'll need to watch only a few sections of music with any kind of dynamics for it to peak out too high if your's is doing it.

I'm using FLAC level 8 and RG set to 89dB for RG'ed and -10 for non-RG'ed. and before anyone asks, yes all the files I'm referring to are RG'ed . If I remove the RG tags, it uses the -10 as expected. Thanks for the suggestions, maybe someone can try this and confirm what I'm seeing.

Replaygain applied files not limited to same volume

Reply #7 –
I get the impression you don't fully understand how replaygain works.
It does nowhere limit the level of the sound. It calculates a gain level that would most of the time give the same loudness impression. For very high dynamic music this is a bit of a problem and although RG usualy does a good job, for some exceptional files it might be off.

And, on a similar subject, replaygain does not always prevents clipping. For modern loud (or loud mastered) music it does, but not always so for high dynamic music. In foobar2000 however, you can select under playback preferences -> Replaygain -> apply and prevent clipping. Maybe that is what you're looking for?
In theory, there is no difference between theory and practice. In practice there is.

Replaygain applied files not limited to same volume

Reply #8 –
Quote from: GeSomeone on
I get the impression you don't fully understand how replaygain works.
It does nowhere limit the level of the sound. It calculates a gain level that would most of the time give the same loudness impression. For very high dynamic music this is a bit of a problem and although RG usualy does a good job, for some exceptional files it might be off.

And, on a similar subject, replaygain does not always prevents clipping. For modern loud (or loud mastered) music it does, but not always so for high dynamic music. In foobar2000 however, you can select under playback preferences -> Replaygain -> apply and prevent clipping. Maybe that is what you're looking for?


OK.. sorry, not "limit" I shouldn't have used that term. I used that  term because for all the music I know I hear as the same volume (because of RG), when I watch the output, none of it goes above that "magic" -10dB volume. I know it is only because of a scalar multiplication of the sound signal by the gain album value (in my case).

What you said about RG "usually does a good job" has me interested.. do you have albums that despite being RG'ed, are still loud enough in volume compared to all your other RG'ed music that you need to turn down the volume (of external amp etc) when you reach that album? I could understand if RG isn't 100% perfect since it's working on "perceived" volume. But intil recently, I've never had to change the volume at all while using RG. And now that I'm finding I have to, it's quite annoying. It's nearly a big enough difference in overall volume that it's nearly not worth bothering with RG and just set the overall foobar volume to -10dB to deal with most of the newer "hot" albums (which is why I started using RG in the first place).

Thanks for the help.. it's just frustrating when it was perfect and now it isn't quite right. And I'm picky, especially when I know it CAN be right

I'll continue fiddling and reading.

Replaygain applied files not limited to same volume

Reply #9 –
what I was trying to get atwas.... in album gain mode, not all tracks are attenuated to the same peak volume (as they are in track gain mode)... so some tracks will be louder and some will be softer... look at a whole albums worth of replay gained files... the album gain value is consistent across all tracks, but track gain levels will be both above and below the album gain level on a track to track basis... so applying the album gain during playback will not produce volume levels that are consistent...

it sounds as though you want a maximum peak gain level for all albums to be the same... yet if your using album mode you'll still have some above (and below) that level... for almost all cases I would think a 5db increase is larger then there should be (so there still may be an issue)... but try looking at the track gain values...here is an example:


above you can see that if I use RG in track mode I will get 89db gain level playback with the track gain value... well if I only apply album gain then some of those tracks (such as "new milenium") will still be approx. 1.25 db over my reference level... and some will be brought lower then 89db (when album gain exceeds track gain)

as I said the 5db+ differences your hearing/seeing seem pretty extreme so there maybe something we're not considering... but look at the track gain levels for the affected tracks and see if there is any correlation there...

I don't have any of the albums you listed at work (though I have the dream theatre ones at home), so I can't see how they act on my setup till I get home on sunday...

Peace
Ross

Replaygain applied files not limited to same volume

Reply #10 –
Thanks, I appreciate you looking into it in detail. I understand everything you've said so far and agree with most. However, I'll show you what I have issue with using images (thanks for the idea!)

First of all, yeah i understand that "applying the album gain during playback will not produce volume levels that are consistent" WITHIN A CD... Correct. But between CDs the maximum absolute volume should be very very close! Until recently, while applying album gain, I was able to set the volume i wanted and forget about it, even after an album change. Now I can't because I'm getting big variations in maximum volume on some *albums*. So don't worry, I know how track/album gain works.

Below is a picture of one of the tracks that is too loud. It was backed up from CD using EAC. Converted, tagged and replaygained with Foobar 0.9.1. Then played. On the left is my E-Mu 0404 soundcard mixer/monitor. I have no effects or anything, simply monitoring WAV and ASIO output. Foobar is only using ASIO. And I'm using album gain.


This is at one of the loudest sections of an Infected Musrhoom song. You can see that the output is WELL above the 10 marker. So much so that if u look closely, u can see the mixer is complaining that the channels are maxed out (from the red lights above). I have to click on those dots for them to go away (it acknkowledges the fact that you've seen they were maxed out at some point). This was from track 03, and the RG values for this album are:



Next album is one that is fine: Strapping Young Lad's self titled album. For this album, the output meter never ever goes above the 10 mark you can see in the image. This is where all my fine albums appear to max out. When there are quieter sections in the music, of course it drops off a little. The albums which are playing with a louder maximum volume do the same in that they have quieter sections. It's just that when they ARE loud, they are at the unacceptable higher limit. I have tried to determine if the bad albums are all reaching the same upper limit or if it's random (ie related to the music, not some common faulty setting). I can't tell so far. Anyway here is an example of a good album and where I want all my music to max out as far as volume, no matter how loud they are (what I want RG to do! To make all my CDs have a common maximum reference volume of 89dB, which from what I can tell, appears to line up exactly with that 10 mark / -10dB on my E-Mu mixer):



Here are the gain values I got for that SYL album:



Now, you might say "well duh, the SYL album is a heavy metal album and has a higher negative adjustment". But, isn't the whole point of those scalar album gain values to bring them all close to the same volume BETWEEN CDs, yet keep the dynamics of each track within each album?
My understanding of RG in album mode is: you get quiet album A from the 80s. It only gets maybe -1dB to have an overall volume of 89dB. Then you scan something new like SYL and it gets -10dB cos it was mastered way too "hot". Now, that -10 is supposed to bring that album to the same (maximum!) volume at it's loudest sections, as the 80's album. And then the same for anything else you scan. If the scalar amount RG comes up with doesn't make the albums have roughly the same overall maximum volume (ie leaving the dynamics WITHIN each cd from track to track), then it isn't working!

As you can see, I'm getting a crazy maximum volume on some albums (where the volume can completely max out the meter, now that I've found a few more than do it), yet on the fine albums, they all max out at that -10dB marker like soldiers! Not a single toe stepped over that line. This is what's completely stumps me, that it can get so many of the albums perfect on that line and I can safely just set the volume I want for the first thing I listen to and be fine for everything that day (actually i used to be able to set once and forget it forever after!). But now I'm getting these albums which appear to refuse to follow the rules. If I remove RG info, these albums basically max out the output meter non-stop as you would expect. So RG *IS* being applied, but I don't think it's correct as for some albums it's toe'ing that line (the way I like), for others it goes a lot higher depending on the volume of the song at the time.

Hopefully you can see what I'm saying now, I don't think I can make it any clearer. And I believe I understand how RG works. I know that if I set it to track mode, all my songs will peak at the same value and it'll sound like the radio. And using album gain as I do now, some tracks are inherently quieter as they are intended to be, while others hit that maximum (which I expect is the 89dB volume aimed for by Foobar's calculations). I believe that the 89dB volume is what Foobar aims to have as your max volume, and it scales the volume of the whole album using the album gain. There will always be variations in the volume depending on the song at the time. It's just that maximum volume that I'm really trying to nut out and figure why some albums are way louder. Both the "playback" images in this post were taken exactly at points in songs where the volume was the loudest for that particular track (and I picked songs that were representative of the whole aulbum, not quiet filler tracks). I know some will argue that it's hard to take the pics correctly to show what I intend, but these two albums both have sections of pretty much "full on volume"... you know, where you can tell it doesn't get any louder than that on the CD. I took pictures at those points.

Phew!

Replaygain applied files not limited to same volume

Reply #11 –
the point of album gain is ensure you don't really need to change the volume much between tracks in a specific group, rather than across groups.

Its more along the lines of saying you want to make a mix cd, and don't want to adjust the volume in your CD player between tracks, or listening to all of an artist's work in order by album release.

in order for foobar to do what you are expecting it to do, it would need to keep the album gain info for all the albums on a given playlist in memory rather than what it does now, loading the RG info on a per-track basis. Or scan all the files in a given playlist as a single album

Replaygain applied files not limited to same volume

Reply #12 –
I'm not familiar with the songs, so I just put in my 2 cents.

RG does try to limit the peak amplitude. That would be normalizing. What it does is try to make the average effective (e.g. RMS) loudness the same.

If a track has very silent portion and very loud portion at the same time, the average effective RMS may be lower than the 89 dB, causing a + value. This may even cause the loud segments to clip! Of course this is somewhat unlikely, but possible.

But then again, it's only my 2 cents

Edit: Whoopsie. A typo. "RG does not try to limit the peak amplitude," I should've written.

Replaygain applied files not limited to same volume

Reply #13 –
OK I've figured out something, but I don't like it...

Quote from: funkyass on
the point of album gain is ensure you don't really need to change the volume much between tracks in a specific group, rather than across groups.

Its more along the lines of saying you want to make a mix cd, and don't want to adjust the volume in your CD player between tracks, or listening to all of an artist's work in order by album release.

I agree with this part if by "groups" you mean albums.

Quote from: pepoluan on
RG does try to limit the peak amplitude. That would be normalizing. What it does is try to make the average effective (e.g. RMS) loudness the same.

I assume this is a typo and you meant to say "RG doesN'T try to limit the peak amplitude". Because you then say what it does (and that part is correct). I'm not sure if you've read all the above, but yeah I admitted I used an ambiguous phrase earlier on. I'm aware of the difference between RG (pure linear scaling of all frequencies) and normalisation (screws with the quieter audio information with respect to the loudest peak) and corrected how I was expressing myself.

Quote from: funkyass on
in order for foobar to do what you are expecting it to do, it would need to keep the album gain info for all the albums on a given playlist in memory rather than what it does now, loading the RG info on a per-track basis. Or scan all the files in a given playlist as a single album

This I disagree with! I don't get why you are saying this when we have already established that the album gain value associated with each album is supposed to scale their maximum volume (and as a result, all the rest of their dynamics AT CD LEVEL, not track level) to the same value selected in Foobar (89dB). Also, there is NO point whatsoever in scanning all tracks from sets of multiple albums, as if you use album gain on this set, you get the same differences between albums that you would get if you didn't use RG at all! And the track gains are exactly the same as if you scanned the albums individually. Because it will analyse and scale all the songs individually according to the quietest song and the hottest songs you have in your collection and calculate to reach the 89dB. However, here is the interesting (annoying to me) part: When scanning multiple albums as ONE album, as suggested by you, I can see that the peak amplitude reached by all the albums is now the same. This is what I wanted to achieve in my soundcard mixer. BUT, the albums which were quiet without RG use now sound definitely quieter than the albums which were louder without RG. That is because it is now calculating the maximum peak (and  therefore the 89dB) according to those loud albums, and the quiet ones all get scalled down accordingly to match, nearly like they were just quiet songs on an album (where it IS the desired result, but here it isn't).

But this MUST be incorrect use of RG for two reasons:
1) The quiet albums STILL simply sound quieter compared to the loud albums. If you're using album gain, I believe that all albums should then have the same (via my ears and via monitoring output levels) max amplitude, and the dynamics of all the tracks within each CD remain, as they are scaled accordingly. But it doesn't. This is incorrect behaviour of RG since as stated by others in this post, it is supposed to make the average loudness of all the albums the same in album gain mode.
2) And this is the main reason I believe scanning multiple albums as a single album to achieve a "better" RG album gain is incorrect: this requires REscanning your entire collection each time you add a new album!!! That is absolutely rubbish and MUST be wrong. The whole point of that 89dB value is so that as you add new albums you can just scan them and the album gain will be based on that value you chose when you set up Foobar.

So yeah, the peak amplitude reach when scanning multiple albums is "perfect" when monitoring the output meters. It never goes above that limit. But more importantly, the sound levels to the ears are definitely inconsistent between albums, much along the lines of if RG wasn't being used at all!

No matter what happens, I don't intend to rescan my entire collection each time I add a new album. So, currently my best solution is that suggested by Lyx. I've set my overall Foobar output to -5dB to stop the albums which STILL appear to max out my soundcards digital output. I've adjusted my amplifiers volume to account for this, and I'll continue to use album gain. The albums which sounded too loud are now sitting around the -10dB when at they're loudest, as desired. The albums which were fine before now LOOK like they are coming out too quiet, but SOUND ok. It annoys me that the output monitor doesn't entirely match up visually with what my ears are hearing. I think someone said before that RG doesn't deal very well with some albums, after more investigation, I've found that in general, albums which have a larger dynamic range are the ones that are coming out with a hotter max volume. Heavy metal and stuff that probably has less dynamics (ie all loud) are the ones that are "well behaved" and don't go above that upper limit I like to see on the peak meter. Problem is, those CDs now sound a bit quieter. EG Linkin Park's Hybrid Theory (on average) sounds quieter than Madonna's Confession on a Dancefloor. This part really irks me. LP has an album gain of -10.x, and Madonna -8.x. I guess I've just gotta put it down to Foobar's RG calculations not being as accurate as I'd like. It's a LOT better than not using RG and albums playing at completely different volumes, I just wish is did it better (and I don't want to go manually adjusting album gains). Unless someone has any new ideas I think this thread as been run into the ground and I'm tired of trying to "fix" it. Thanks to all who helped me.

Replaygain applied files not limited to same volume

Reply #14 –
Quote
If you're using album gain, I believe that all albums should then have the same (via my ears and via monitoring output levels) max amplitude

That would be peak normalisation, not Replay Gain. And no, normalisation is linear scaling as well; maybe you're confusing it with dynamics compression?

Quote from: sl33py on
Correct me if I'm wrong, but isn't the whole aim of using replaygain "album gain" to keep all the dynamics within the albums intact, but for the maximum volume of all albums to be scaled to the same absolute volume

You're wrong, and I'm surprised nobody seems to have pointed this out. (This thread is full of text, and difficult to read, so I might be wrong, but if it's there, I'm still surprised it isn't very noticable.) Peak amplitude says nothing about perceived volume.

Replay Gain normalises perceived volume, not peak amplitude. If you want peak amplitude to be constant, use maximum pre-amp for RG-scanned files and scale down tracks that would clip. The downshot of this is that the perceived volume *will* fluctuate, and you'll have to reach for the volume control.

I would urge you to stop looking at the VU meter. Instead, you should listen to the audio.

Replaygain applied files not limited to same volume

Reply #15 –
Quote from: sl33py on
Quote from: rosshmusic on

are you using "track gain"... or "album gain" mode for playback...?

with track gain I would expect them to remain fairly consistent... but with album gain there could be differences...


Album gain. But I disagree with you. The albums should have the same *peak* volume using album gain RG. If you turn off RG and just set foobar's volume to -10dB (on tracks with and without RG), that would get rid of the "hotness" from newer albums and clean them up a bit, but it would also make the "good" old albums much quieter at the same time.


You're thinking of peak normalization.  Replaygain puts very little restriction on the peak value, since the peak value often does not correlate with percieved loudness.

Edit:  And it helps to be concise when you want free tech support.  Judging by how no one until the last couple posts even noticed that your premise was off, I can only assume people weren't actually reading through all those huge blocks of text.

Replaygain applied files not limited to same volume

Reply #16 –
Granted @ peak normalisation. I think someone else brought up normalization, when I avoided using the term because I know it's not what I was trying to achieve. And yes Mike Giacomelli, I did actually rephrase myself when it was pointed out my firsts terms were incorrect. But yes I think one or two ppl didn't read the whole thread before answering. Maybe because I was too verbose initially, but I was just trying to make myself understood *runs away*

Even now that I'm ignoring the output meter, I'm just having trouble believeinf that I can still tell a noticeable difference between albums in what I perceive with my ears, which is what RG is supposed to correct/alter. I thought the albums would all be much closer to the 89dB "perception" it aims for, with none of these "rogue" albums I have. Thanks all.

Replaygain applied files not limited to same volume

Reply #17 –
1. Perception is a tricky matter, which is why we have TOS #8. You might have been tricked into thinking that some files are louder, when really they aren't. Lookinig at that VU meter might have been a mistake.

2. I think I remember reading that RG calculations can be off by +/- 1 dB.

Replaygain applied files not limited to same volume

Reply #18 –
Quote from: sl33py on
Quote from: pepoluan on
RG does try to limit the peak amplitude. That would be normalizing. What it does is try to make the average effective (e.g. RMS) loudness the same.

I assume this is a typo and you meant to say "RG doesN'T try to limit the peak amplitude". Because you then say what it does (and that part is correct).
Yeah, my bad. Edited my posting.

Quote from: sl33py on
So yeah, the peak amplitude reach when scanning multiple albums is "perfect" when monitoring the output meters. It never goes above that limit. But more importantly, the sound levels to the ears are definitely inconsistent between albums, much along the lines of if RG wasn't being used at all!
May I offer a hypothesis?

It might be the case like I posted above, i.e. a track/album with an extremely loud segment, with many quiet segments around it. The RG value will then be +. But fb2k tries to prevent clipping, so it does not apply RG. Try instructing fb2k to not prevent clipping. If my hypothesis is right, your albums will have same perceived loudness. Badly distorted due to clipping, but same loudness.

You're f***ed up there, I'm afraid

Replaygain applied files not limited to same volume

Reply #19 –
Quote from: pepoluan on
It might be the case like I posted above, i.e. a track/album with an extremely loud segment, with many quiet segments around it. The RG value will then be +. But fb2k tries to prevent clipping, so it does not apply RG. Try instructing fb2k to not prevent clipping. If my hypothesis is right, your albums will have same perceived loudness. Badly distorted due to clipping, but same loudness.


I've played with the gain settings so much it's not funny. Using prevent clipping or not had no effect that I could hear.

RE track with very loud segments, then quiet ones around it, this is happening on multiple albums in songs with good mastering (ie lots of dynamics and varying volume) and also in songs where it's just full on blasting (ie poorly mastered heavy metal). It's not a function of song volume.

I wouldn't be worried if the volume variances were +/- 1dB. But they are sometimes in the vicinity of up to 5-7dB, enough to make me reach for volume.

Replaygain applied files not limited to same volume

Reply #20 –
What if you use track gain on playback?

Quote from: sl33py on
Track gain should not be used on albums at all, only on single tracks where it is impossible to calculate how "loud" they should be compared to the other tracks on the album they came from.

There's your trouble, I think.

With album gain, a single quiet track from a generally loud album would sound quieter, and a single loud track from a generally quiet album would sound louder.

Also, RG analysis does not calculate how loud songs (or albums) should be, compared to other tracks; it calculates how loud songs (or albums) currently are, compared to a fixed reference level.

Replaygain applied files not limited to same volume

Reply #21 –
Quote from: kjoonlee on
What if you use track gain on playback?


Then I'll lose all my dynamics between tracks when listening to a whole album (as I usually do). I don't intend on doing that. Soft songs will sound as loud as the loud songs. Track gain is not the solution I'm after. I already know that. To satisfy curiosity though, I played back the albums I've been tesing using track gain, and the obvious volume difference still remains! All songs from the 'problem' albums are louder than the other albums.

The following is only true if you add the extra words in brackets. But anyway, I'm talking about loud tracks from different albums, so it doesn't really help figure out this problem:

Quote from: kjoonlee on
With album gain, a single quiet track from a generally loud album would sound quieter (compared to non-RG playback), and a single loud track from a generally quiet album would sound louder (compared to non-RG playback).


If you were to take the loudest songs from two different albums (doesn't matter if the albums are loud or quiet overall), and play them both using album gain, they should sound very close to being the same volume at their loudest points, because RG is meant to make them sound very close (to your ears, regardless of output meters etc). But I obviously get the loudest songs from multiple albums sounding like they have very different volume levels (definitely maximums but also in general). This is the reason I use overall loud tracks from different albums when testing for this volume difference. There's no point testing moderate or quieter songs if I want to see where the volume is peaking out for that particular album.

Quote from: kjoonlee on
Also, RG analysis does not calculate how loud songs (or albums) should be, compared to other tracks; it calculates how loud songs (or albums) currently are, compared to a fixed reference level.


This I also know. It goes "ok this cd from the 80's has a maximum of about 80dBm so it's a bit quiet and this new heavy metal album has a maximum of 100dB, so when I playback the 80's cd, I'll add 9dB to bring it up to the reference level the user has chosen (89), and with the heavy metal album I'll subtract 11dB from it's volume so it also sounds about the volume the user wants.

When some people on here talk about RG affecting perceived volume level, it nearly sounds like they think it's doing all kinds of magical voodoo stuff with normalisation and junk.. it's not! In album mode, all it's doing is multiplying the volume of the original media by a pre-calculated factor to make it playback at the same volume as every other album in your collection. Because it is simply a volume adjustment, when I apply album mode, all the dynamics remain, and I expect the maximum volume of all my albums to peak out at *roughly* the same volume. How else is it going to stop u reaching for the volume between albums? It can't. The other reason I expect the maximum volume to peak at the same place for all albums is because on 95% of my albums that I'm happ with, it does! And this is why when some albums are visually and audibly being played back a good 5-7dB louder than the rest, there is an obvious problem.

I wouldn't really worry about this problem any more though, several people seem keen on telling me I'm using RG incorrectly, or expecting the wrong kind of behaviour from it. I refute this, because I'm comparing the perfect behaviour of about 200 odd albums, with about 10-20 which behave differently. I believe the 200 albums are fine and they playback as I want and expect, no reaching for the volume whatsover, and these include, old and new music, across all genres. So I use them as reference for the problem albums (which also vary in release date/genre). If it was more like 50/50 then I would definitely question how I was using RG. But right now, I dont.

I'm just going to go in and use my ears and adjust the RG album gain on those albums that I hear as being too loud at their loudest points.

Replaygain applied files not limited to same volume

Reply #22 –
Quote from: sl33py on
Quote from: kjoonlee on

What if you use track gain on playback?


Then I'll lose all my dynamics between tracks when listening to a whole album (as I usually do). I don't intend on doing that. Soft songs will sound as loud as the loud songs. Track gain is not the solution I'm after. I already know that. To satisfy curiosity though, I played back the albums I've been tesing using track gain, and the obvious volume difference still remains! All songs from the 'problem' albums are louder than the other albums.

That was not a suggestion per se, but an attempt at further diagnosis. Anyway, you're still being vague. How did you measure loudness? By measuring maximum loudness?

Quote from: sl33py on
The following is only true if you add the extra words in brackets. But anyway, I'm talking about loud tracks from different albums, so it doesn't really help figure out this problem:

Quote from: kjoonlee on

With album gain, a single quiet track from a generally loud album would sound quieter (compared to non-RG playback), and a single loud track from a generally quiet album would sound louder (compared to non-RG playback).


Not really. It's still true without *your* added parts.

Quote from: sl33py on
If you were to take the loudest songs from two different albums (doesn't matter if the albums are loud or quiet overall), and play them both using album gain, they should sound very close to being the same volume at their loudest points, because RG is meant to make them sound very close (to your ears, regardless of output meters etc). But I obviously get the loudest songs from multiple albums sounding like they have very different volume levels (definitely maximums but also in general). This is the reason I use overall loud tracks from different albums when testing for this volume difference. There's no point testing moderate or quieter songs if I want to see where the volume is peaking out for that particular album.

No, you should abandon your thoughts about loudest points, because RG is not peak normalisation. You've been told this over and over.

Quote from: sl33py on
Quote from: kjoonlee on

Also, RG analysis does not calculate how loud songs (or albums) should be, compared to other tracks; it calculates how loud songs (or albums) currently are, compared to a fixed reference level.


This I also know. It goes "ok this cd from the 80's has a maximum of about 80dBm so it's a bit quiet and this new heavy metal album has a maximum of 100dB, so when I playback the 80's cd, I'll add 9dB to bring it up to the reference level the user has chosen (89), and with the heavy metal album I'll subtract 11dB from it's volume so it also sounds about the volume the user wants.

RG is *not* about maximum volume.

Quote from: sl33py on
When some people on here talk about RG affecting perceived volume level, it nearly sounds like they think it's doing all kinds of magical voodoo stuff with normalisation and junk.. it's not! In album mode, all it's doing is multiplying the volume of the original media by a pre-calculated factor to make it playback at the same volume as every other album in your collection. Because it is simply a volume adjustment, when I apply album mode, all the dynamics remain, and I expect the maximum volume of all my albums to peak out at *roughly* the same volume. How else is it going to stop u reaching for the volume between albums? It can't. The other reason I expect the maximum volume to peak at the same place for all albums is because on 95% of my albums that I'm happ with, it does! And this is why when some albums are visually and audibly being played back a good 5-7dB louder than the rest, there is an obvious problem.

RG is not about maximum volume.

Quote from: sl33py on
I wouldn't really worry about this problem any more though, several people seem keen on telling me I'm using RG incorrectly, or expecting the wrong kind of behaviour from it. I refute this, because I'm comparing the perfect behaviour of about 200 odd albums, with about 10-20 which behave differently. I believe the 200 albums are fine and they playback as I want and expect, no reaching for the volume whatsover, and these include, old and new music, across all genres. So I use them as reference for the problem albums (which also vary in release date/genre). If it was more like 50/50 then I would definitely question how I was using RG. But right now, I dont.

I'm just going to go in and use my ears and adjust the RG album gain on those albums that I hear as being too loud at their loudest points.

RG is not about maximum volume.

What is RG about? It's about *average* volume.

Replaygain applied files not limited to same volume

Reply #23 –
Wow, there's an echo in here...

I listened through my collection and found that there are significant differences in the average volume I perceive across many albums. A minority of albums are too loud and I need to turn down the volume because the difference is loud enough to hurt the ears, while the others are perfectly acceptable. These are facts, regardless of anything else. Therefore I've come to the conclusion that Foobar either doesn't calculate or implement Replaygain information as well as it intends to, to reach a common average volume.

NOW, as far as replaygain NOT affecting maximum volume. I would like you to slowly and carefully read this.

Consider: What operation do you think Foobar performs on the audio files, considering it has only the audio you give it, and a single scalar RG number (whether it's track OR album gain). It SCALES it.

Therefore, when Foobar scales the output by the scalar RG value calculated earlier, everything is either shifted up or down in volume.

This Includes:
1) the quiet stuff... minimum perceptible volume
2) the moderate stuff... everything that isn't minimum or maximum volume
and
3) MAXIMUM VOLUME. RE THE LOUDEST STUFF

I appreciate that you have tried to help kjoonlee, but don't treat me as an idiot when it is you who has missed what I have now said repeatedly. Never did I say that RG affects only maximum volume. A high post count does not give you any right to talk down to people.

I have no more to say on the matter. As far as I care, this thread can be locked/closed/disintegrated, as it didn't help me and won't help anyone else who hears what I hear.

Replaygain applied files not limited to same volume

Reply #24 –
/me shrugs.

I have never ever thought you said that RG affects only maximum volume. You had missed my point, which was that RG is not based on maximum volume.