This wavform is not alter by any way. Taken directly from lossless file ripped from CD.



As you can see, the track is clip badly but for some reason there's a quite huge headroom left (FB2K RG scan peak value = 0.793701, Track gain = -9.73db) which I thought it can be use (when mastering) to "expand" the waveform a bit to make it not as badly clip as it is. So my question arise.. Why they don't use it? Is it more beneficial to leave a headroom and then just clip the track instead of using it?

ReplayGain is attempting to match the relative loudness of the track to the 89dB SPL reference level. It will almost always leave some headroom regardless of whether a particular track needs it or not. That's what ReplayGain is designed to do.
It's too late. The damage has already been done at an earlier stage by an engineer not giving the signal any headroom. ReplayGain can't undo clipping on the original source material.
No. It looks as though whoever mastered the original recording doesn't understand the concept of headroom, hence the source material already being severely clipped. Sadly, it can't be magically undone.

Cheers, Slipstreem.

That's not a track from (What's the Story) Morning Glory is it? Mastering was the same on that CD. -9.00 album gain with a max peak at 78% So normalizing would have made it almost -12.00.. a record from 1995 mind you.

Gillan's Inn by Ian Gillan (2006)

Rectangular music.

There is absolutely no way you can tell from that image that there is clipping. If you don't believe me, generate a 10kHz sine wave with a duration or 30 seconds or so and see what it looks like.

Correct. It might have been aggressively limited, but the only way you can discern full scale clipping is to look for any instance of at least 3 samples of the same amplitude at either amplitudinal extreme, and that's a minimum metric. If you find an instance of three or five sequential samples at the same amplitude, then you can call it clipped (at least that's how it works in "my book").

As for the headroom issue, I recall reading about a time when mastering engineers feared the characteristics of early-ish DACs with respect to handling signals that hit 0dBFS. So, you'll come across a number of CDs that may be limited to -0.1 or even -0.01dBFS and probably a handful that don't pass -1dBFS. It's possible this ME was so fearful about it that he didn't even consider passing the 80% threshold, which is really quite stupid from a perspective of maximizing dynamic range and effectively utilizing available bit depth.

Sometimes, there is indeed a reason to master a track like that. A good example is The Island OST. The last song "Blow" has obviously been mastered by someone else and has thus been louder than anything else on this album. So to correct out this loudness difference, this particular track has been attenuated -- while it looks strange in a wave editor, it has definitely been the right thing to do because otherwise you would have had to pull down listening volume just for this one track.

I don't see that at all. It's hard-limited, but I'd bet money that it hasn't been clipped. They're very different things. As with many on the board, you're looking at a picture of audio and assuming you can draw conclusions about its sound.

That said, if you wanted your audio to visually "look" better, you could run it into iZotope's Advanced RX application and see what it did to "repair" it in the declipper. I think it's mostly a mind-game, but if it helps you for your audio to look all curvy, then you can always play with it.

CODE

http://www.izotope.com/products/audio/rx/

@Raiden







Enough?

The idea was, Since my question is "Why did they leave the headroom there and have sqaure shape waveform instead of using them" I'm not trying to "restore" it or complain or whatever about it. So, I don't want to put up too much image on the board and show just the overall average of the waveform instead of zooming it like someone trying to find the fault in the music or lossy encoder and I don't see why it is need. So I naively taken for grant that people would believe my word and answer my question instead of questioning me.


Now you can, and yes it's clipped, so I want my money.

@Slipstreem

I didn't expected RG to do anything, I just show the value of the peak & RG value just for information sake.

I think I can answer these question myself (and how can I never thought about it before posting..) This one is come from a CD that kinda like "HIT" CD (it's not, but from various artist anyway) so the "original" may already clip and when they put it together with other track from another artist they just lower the track amplitude to match overall loudness. (Same answer as Gasmann, if I don't figure this out by myself first this should answer my question.)

I'd rather question you than have people get the wrong impression from what you wrote.

I've learned my lesson, I should just said "The track is clip but peak is only at 0.79" with no pic about it whatsoever in the first place. And people will not going to ask me "You sure it's clip?" "Already ABX?" "I bet it's not clip" and instead answer my simple question. (with simple answer too, it just I didn't thought about it before)

Thank you.

EDIT:Typo

Hi, buktore!

It's a perfectly valid question and I also fail to understand why some keep picking on you. Unfortunately I don't have an answer to your question. It just seems to be horribly mastered. There's no good reason for this hard limiting, IMHO. At least none that I know of.

Cheers,
SG

So, essentially, this is what a ReplayGain-enabled-player does when processing output, correct?

But only if you're using track gain. But then again you REALLY don't want to listen to a soundtrack with track gain, do you? Track gain kills inter-track dynamics. It's intended and a really pleasant thing that a loud song has say -3dB and a quiet song +10dB replay gain value. Now if you track gain them, the one with full orchestra playing sounds as loud as maybe a single flute playing. That pretty much ruins everything. That's why I, for myself, always use album gain. That applies one and the same gain to all tracks. Now however, if one track were much louder than everything else, it would also be like that after album gain, because album gain assumes the mastering engineer knows what he is doing (and as I just said, some tracks do have to be louder).

Back on topic, what I wanted to say is that there REALLY is a reason to master a track like that. In this particular case, it was the best decision the mastering engineer of the soundtrack (not of the Blow song, of course!) could do. If a song comes out of a different mastering and doesn't fit because it's too loud, it has to be made quiter. If one song is already at peak 1 and too quiet, all other tracks should be made quiter. As long as peak 1 (or near one) is reached at least once in the entire album, everything is fine and no bits or whatsoever are "wasted". That approach was a standard in CD-Audio mastering before the evil loudness war has arisen.

So @buktore: Is peak 1 reached in the album at least once? If so, it is fine. I also once had a really nice compilation cd (I can't find it anymore ) where some tracks looked like the one you've shown while others were more dynamic and needed higher peaks.

Whatever you say, I prefer that kind of mastering to the modern "every track has to be as loud as the loudest one"-style.

edit: only language issues

Don't worry mate, I use album gain for my album listening.

I was just trying to understand the concept; so the answer to my question is "yes". Thanks.

For a quantized (i.e. not float) lossless format, yes (because there is no content above the valid range).
For a lossy format (or float lossless), the reduction of volume should happen before quantizing, which could imply that content that would have been clipped, it is not afterwards.

This is not something unusual. lossy encoding signals, due to the removal of some frequencies (or addition of noise), generate signals that are different than the original, so the peak could be above the valid range.