I haven't been on these boards for quite a while. After a bit of a hiatus, I've found that on the topic of CD loudness, the gripes are the same. But that's a good thing! It means we're getting more and more frustrated with how loud the music companies are mastering CDs.

Now, in my opinion, some people can do it tastefully (Metallica's "S&M" is pretty loud, but still pretty clear, which is no small feat considering how much is going on in that album), but more often than not you're seeing it done rather horribly. I just got to a thread in this section showcasing a poor mastering of the Corrs' new album. I took a listen to it myself and my jaw dropped. It was awful!

But there IS some hope. I just recently picked up the 2003 remaster of The Police's "Synchronicity," fully expecting it to be either clipped in places, very obviously compressed, or just otherwise made uncomfortably loud.

Upon first listen, it was clear, not clipped, and dynamically very strong! I had to see for myself what the wav looked like. As you can see in this picture, the wav doesn't look like it's being squashed. As far as I can tell, there's some very slight hard limiting, but not much. Also, take notice of the Average RMS reading. Most modern-day albums have a reading around -10. This is significantly lower. (By the way, Audible!, in the Iggy Pop sample you posted up a while back, the Average RMS reading was -3.87!)

In the next post, I'll give you all a FLAC sample of Synchronicity I from this master.

This is 1:00-1:15 of Synchronicity I. About 2.5 seconds into the sample is the peak where it's most evident that they used some hard limiting. Once again, though, it's not too bad when compared to other modern remasters (like DSotM, which is one of the main reasons I signed on to this board in the first place ).

And if anyone has another version of Synchronicity they'd like to compare this to, feel free to share. Actually, if anyone has the MFSL version, I'd love to hear what kind of job they did.

Sounds like a glitch at ~9.8 seconds to me.

edit: whehter or not that's bad mastering or problems with the masters, I don't know.

The waveform looks somewhat quiet indeed. CEP's statistics are not that good for determining a track's loudness, could you offer replay/wavegain album/trackgains ?
I'm going to listen to the sample.
How loud/clipressed are remasters generally ? "remaster" is a flashing warning sign for me ("don't touch !"), thus I can't tell. Are remasters generally quieter than new-mastered CDs ? (CDs from the same year)

In my experience, remasters get louder, but never as loud as the current "new album" standard (-9, -10 replaygain values).

Bob Dylan's Nashville Skyline went from something like +2 db for the original CD to -4.90 db for the most remaster. Ring Them Bells from Oh Mercy went from a trackgain of around 0 db to -5.11 db. There's some light limiting on them, but nothing that really interferes with my enjoyment of the music.

The Synchronicity sample has a -5.16 RG for a loud passage. Pretty dynamic!

For my purposes, CEP's RMS readings are very good in helping me determine loudness. But as far as replaygain numbers are concerned, the first track (Synchronicity I) is at -4.72. Not too shabby!

I haven't listened to too many new remasters just yet. One recent remaster that has me scratching my head is Weezer's debut album from 1994. They've repackaged it as a Deluxe Edition and apparently has been "digitally remastered." First of all, how badly did it really need to be remastered? It's 10 years old! If I had to guess, they probably clipressed it to hell. The sound of the original album lends itself well to a remastering treatment like that.

As I mentioned earlier, one of the reasons I signed on was because this was the one messageboard that actually had statistical evidence that the 2003 remaster of Dark Side of the Moon was made louder (something like 3 to 4dB louder). When I learned about this, I set out to find the "best" mastering of it. That's when I discovered a problem with the 1992 remaster. See this thread for more of that:
http://www.hydrogenaudio.org/forums/index....showtopic=22493

New remasters in general seem to be kind of a gamble. Sometimes they're done right, sometimes not. It does make me wonder sometimes just how involved the artists themselves are in the mastering process. I know if I had a record contract, I'd want to be able to oversee every step of processing my music was going through.

Replaygain Update: Album gain is an impressive -4.54dB.

Synchronicity I: -4.72
Walking In Your Footsteps: -2.64
O My God: -3.69
Mother: -5.13
Miss Gradenko: -3.95
Synchronicity II: -5.51
Every Breath You Take: -5.16
King Of Pain: -4.86
Wrapped Around Your Finger: -3.60
Tea In The Sahara: +0.66
Murder By Numbers: -2.05

I have the MFSL Ultradisc II of Synchronicity and here are the replaygain numbers I have:

Album Gain: -0.45dB

Track gains:
Synchronicity I: -0.88dB
Walking in Your Footsteps: +1.19dB
O My God: +0.11dB
Mother: -2.15dB
Miss Gradenko: -0.74dB
Synchronicity II: -0.21dB
Every Breath You Take: -0.93dB
King of Pain: -0.61dB
Wrapped Around Your Finger: +1.40dB
Tea in the Sahara: +4.89dB
Murder by Numbers: +1.48dB

This may be irrelevant but I have the MFSL issue of Nirvana's Nevermind (the only MFSL I have...scored it in a used bin for $10) and the album RG value is -3.31 dB.

Thread about the Deluxe Blue Album here.

Did you ever get comaprison samples from the CD and SACD layers of the DSotM SACD? If not, I can supply them.

Alsoc, could someone explain to me how to generate replaygain stats for FLAC files, like the ones being shown here?

Foobar2000 comes with a replaygain scanner. Load up the file, right click, scan replaygain.

Yay, people remember all my old threads. =D

The Weezer remaster isn't significantly louder than the original, but the original was up at -9 already, so it's hard to push any louder than that. Interestingly enough, though, if you compare them without using replaygain, the guitars and drums are the same volume side-by-side, but the vocals and guitar solos are mixed much more out front. That's the only reason the album has a louder RG.

Listened to the sample now. (I have to take every sample home in order to listen to it, so it takes at least one day)
It sounds good to me, except the single drumbeat near the beginning, which should have more impact I think. Probably what sounds good is only what is still left from the well-sounding old 80s version and hasn't been destroyed by too excessive volume increase.

What do they do during 'remastering', beside increasing the volume/clipressing ? (Or what should they do ?)
In this thread I compared 2 different mixdowns/masters from the same single tracks. The situation is somewhat similar.
I thiink they apply an EQ curve to boost the bass and the heights, which suffer from long storage on analog tapes. Usually they boost much more than they should. Further they may reduce the noise, maybe this step is done properly.
On a second example I didn't offer in the above thread they reduced the stereo field, made it smaller. The smaller field sounded inferior to me.

Those are impressive numbers. Could you give us 1:00-1:15 from Synchronicity I off the MFSL version?



That IS pretty significant, considering even the original (1991) Geffen release is at -4.98.



I did, actually. The CD layer (and, I can assume, the SACD layer) didn't have the problem that the 1992 one did, which led me to questions I still haven't figured out the answers to: What went wrong in 1992? Was it really taken from the master tapes? Did EMI/Capitol maybe get a second-generation copy of the remaster to put onto CD?



So, technically, the album was not only remastered, it was reMIXED as well. Interesting.



I saw that thread earlier. It was interesting to see the rather big difference between the two masters.

I don't know how much treating they do as far as EQ goes. A purist remastering engineer would apply very little, if any. But here's another example: I have pretty much every R.E.M. album known to man. When their Best Of came out last year, I picked it up for the b-sides and new songs.

But I also was interested in how they treated the older songs to put on their release. When comparing the Best Of's version of Orange Crush to the original one off of Green, there's noticeable clipression, and I even ran Voxengo's CurveEQ on the tracks to compare EQ curves. As it turned out, the highs were actually REDUCED on the Best Of's version. I'm not sure if this is due to simple tape degradation, or because many CDs from the 80s (Green = 1988) had a slightly boosted high end to begin with, and this reduction was how the master tape was "supposed" to sound. Though, sometimes, a slight high-end reduction can also be used as a very simple noise reduction tool.

That fits the other experiences mentioned.
"Remastering" actually has nothing to do with "back to the original", like it should, it's "straight forward to new style". Given the intention to master in a more 'trendy' way, it's obvious that they achieve more loudness by additionally remixing the single tracks. Doing so prooves their evil goals, because for only true remastering it's enough to take the old master tape and remove the damage caused by long-time storage.




I have a lossless copy of their most reknown album "Automatic for the people" from 1994. It is relatively quiet, a bit clipping and compression, but still well-sounding. Could you upload a 30s sample from the greatest hits album from "Drive" or "Everybody hurts" for comparison ? (I suppose they are on the CD.)

There are a few samples hitting zero, but it's never usually more than two or three consecutive samples. And, not to nitpick, but AftP is actually from 1992. Their 1994 album was Monster, and THAT album definitely is a clipressed one.

After looking a little closer, it seems as though the Best Of is compressed/limited, but not really clipped (i.e. No flat-top wavs that weren't there before). Drive isn't on the Best Of (but a live, funkier version is on the b-sides disc), so I included 3:30-3:45 of The Sidewinder Sleeps Tonite.

Side note: I replaygained both the AftP version of Sidewinder and the Best Of version. -6.97 for the original, and -9.21 for the Best Of version.

Thanks tacman,
but I don't have the whole album-if my mind doesn't fail me, I didn't rip "The sidewinder sleeps tonite".
I mentioned "Drive" and "Everybody hurts", because these two ballads are very dynamical on Aftp (especially RMS-dynamics). "Everybody hurts" sounds 'clear' or 'brilliant' on the quiet parts, too, so I thought it'd be good for comparison.
The clipression on Aftp isn't obvious, yes, but in the waveforms there are this areas (vertically) with a high density of peaks-this can't be natural; if you know what I mean. Perhaps it's analog clipression.
I'll analyse your sample, though.
[offtopic]Didn't you have problems with uploading ? I currently can't. [/offtopic]

analog limiters don't have lookahead (obviously). a lot simply clip during the very short attack time, and release in the same way a lookahead limiter would. surprisingly these methods sound much the same, so long as you don't drive things too far.

clipping of very short duration is pretty much inaudible, and doesn't irk me as much as complete removal of a song's dynamics (this is tantamount to removing an instrument in the songs, really - both are included to make the song more powerful and impacting, and the removal of dynamics is nothing less than a perversion of the artist's intentions, unless they actually intend to write boring songs...).

i believe the BBC did listening tests back in the day on tolerable amounts of clipping. IIRC flat-tops of about 6-10ms are not audible in most cases. of course, if the 6ms clips happen on every peak they'll become audible.

Ah, is this the way clipression like on my avatar is caused ? For the first ten samples or so a really flat (clipped) line and then the line falls down as a round curve (or as a straight line) ?
This real clipression, combination of clipping and compression, is it what can be found on most modern CDs, pure perfect flattops are very rare.

Uploading is very limited for me. I can only download 15-second clips. If I try going over that, it says that the upload limit's been exceeded.

Here's a section of Everybody Hurts from the Best Of that takes from both the soft and loud parts of the song. It's roughly 4:00-4:15.

I compared the EH samples. (I chose the appropriate 15s from my copy.)
The wavegain difference is about 3dB. After synchronizing, I could make an ABX test on a loud drum/percussion beat. I ABXed 11/11.
There seem to be no obvious differences, so no remixing or remastering here. I wavegained and inverted the best of sample and mixpasted it over the original. I could clearly see a few parts with loud pops then, they are exactly at the positions of the loudests peaks in the original. So obviously they used some sort of hard limiting, while also amplifying the track by 3dB. They don't seem to have applied clipping. (yes, for stupid engineers clipping is a valuable effect)