Some observations on limiting for loudness…

[miroslav]

When I recently mastered my album, I did three EQ versions of each song with slight variations here and there, and then for each EQ version I did three different “limiting for loudness” versions, using light, medium & heavy settings relatively…though none were “nuked” even with the "heavy" limiting!

While the heavy limited version was certainly commercially competitive, and actually sounded pretty good…I wanted to compare the three limited versions to see what kind of effect the limiting had on the actual sound, not just the loudness.
So…I lowered the output level in the medium and heavy versions so that they had a similar overall “loudness” as the lightly limited version…and then did a bunch of A/B comparisons for different songs and at different sections of each song.

What I noticed on every song was that the lightly limited versions had an bit more “air”…openness, while the heavy limited versions sounded just a pinch stuffier. It was a very mild difference, but noticeable.
The overall difference in loudness is 3.75 dB…that’s how much I dropped the “heavy” limited version to make it sound comparable to the “light” version.

While I know everyone wants competitive loudness, it’s certainly something to think about. I was getting ready to send my final version out for replication…and while I originally was going to send the “heavy” limited version, after doing the comparison, I’m going to sacrifice the 3.75 dB in greater loudness and go with the “light” version, and just let the listener turn up the volume instead.

 

[Massive Master]

Welcome to my world...

Now if we can only get *all* the artists and labels to come to that conclusion (because the listening public certainly never asked for it).

 

[SouthSIDE Glen]

What I noticed on every song was that the lightly limited versions had an bit more “air”…openness, while the heavy limited versions sounded just a pinch stuffier. It was a very mild difference, but noticeable.

A lot depends upon the music and content style, but also one can often hear that things are "pushed". This can be especially true on more ballad-style songs and wall-of-sound-free arrangements. "Pushed", in my own vernacular, refers to an audible loss of dynamics to the point where you can just plain hear that the volume envelopes are unnatural. There no "distortion" in the common sense, in that everything sounds tonally correct and clean, there's no noise or pumping or breathing or anything like that; but the natural dynamics that one ears recognize in real life have been so removed that their loss just bugs you like an itch that won't scratch.

I’m going to sacrifice the 3.75 dB in greater loudness and go with the “light” version, and just let the listener turn up the volume instead.

This deserves recognition and applause; most people don't have the testicular fortitude to believe their ears and follow reality over the peer pressure of a bunch of strangers with the ears and skills of The Who's Tommy .

G.

 

[miroslav]

I know this is obvious to you guys but what happens with most people when they compare the heavier limited versions to unlimited or lightly limited ones is that they probably DON'T adjust the overall output volume during comparisons...so, the heavier limited versions are much louder and initially sound "better".
The unlimited or lightly limited versions almost sound lame after you hear all that loudness.
And that's the trap.

But...lower the output volume (or raise it on the lightly limited versions) so that they are all fairly equal in loudness...and I think most people with fairly well trained ears would notice the more open/airy quality of the lightly limited versions.
The way heavier limiting pushes things up and literally "stuffs" it, it gives the overall sound a somewhat boxier quality.

Frankly…my original mixes without any final limiting on the stereo mix sound the best (when I turn up the volume on the monitors)…BUT…they are also much lower in overall final loudness, so I didn’t want to have a HUGE difference in level when compared to commercial releases.
I think my light limiting is a good compromise...it certainly punches up the loudness, but without the boxy/stuffy quality, and certainly none of that harsh/flat-topped crap I hear on so many “nuked” releases.
I’ve heard some stuff that is literally irritating to listen to!!!
It makes me wonder what the artists/producer/engineer were hearing when they put it out…???

 

[Seafroggys]

Hey, Tommy could blow all those pinball champions out of the water!

 

[WhiteStrat]

I did this same thing last year with an album I recorded for a band of teenagers. They had a little wall-of-sound stuff that could've handled some serious limiting, but mostly they were jangly pop/punk.

When comparing their rough, faders-up mixes to some commercial reference CDs (they were educating me on their target sound! ), they commented on how quiet theirs sounded (they were real happy up until this point).

I assured them that this would be dealt with later, and we went about finishing recording. A few weeks later, when the mixes were complete, I had them all into the studio and let them listen to three mixes: their "un-mastered" mixes, a second set with some manually fixed peaks and moderate limiting, and a third set smashed enough to match the commercial CD.

They were real receptive to a little education on the process, and in no time at all voted unanimously to go with the moderately limited versions.

(BTW--manually scrubbing mixes for errant peaks--usually a drum hit of one type or another--will lead to much better results with the limiter at the end of the game.)

 

[WhiteStrat]

More thoughts on limiting...

For me, the biggest part of "mastering" (yes, I master my own mixes) is making all the songs sound like they belong together. In working towards this goal, I have learned that content (from song to song) limits my ability to limit (pun intended).

Case in point: Right now I'm working on a double CD with 23 songs. A few are as simple as a voice (female) and grand piano. Others are full-out band numbers: drums, bass, keys, acoustic, and electric.

Now with a nice even piano/vocal ballad, there's a ton of headroom, and I can make it LOUD without even killing all the dynamics. But when the very next song is a full band song, the two don't "match" at all if I squash 'em both to the same volume level.

In the first one, her voice is half the mix. Listen to that for three minutes, and your mind forms a picture of how "big" the mix is. But in the full band song, her voice is but one out of 5 or 6 elements (loud ones at that). So even though song two is technically just as loud as song one--it actually sounds smaller in spite of all the other instruments. The mind focuses on the singer as a reference point, and suddenly, in a bigger mix, she's gotten smaller/quieter.

So I get all the mixes done, and apply my moderate limiting to the fullest, most robust mix. Then I work down from there--from big & complex, to small & simple--always making sure the singer stays the same size.

Then when I check out the mixes in my car or whatever, even though the tracks are moving up and down in terms of real volume, they're very natural in terms of perceived volume.

 

[Seafroggys]

Very much the same thought processes I went through when "mastering" my CD.

I lined up all the songs in a single session file, and set the faders so I had the relative volume I wanted. Then I did a test render to see how loud I could normalize it. I went back to the original mixes and scrubbed out the big peaks, which were usually snare hits that happened to have been twice as loud as the other snare hits, and by using volume automation I could turn that significantly down without any difference in the sound. Rinse and repeat until in the master session, those snare hits usually were hitting at -2 to -1, without any limiting at all.

In fact, I didn't even limit or compress, I just used TesslaPro (a transient smoother I like to call it) and got pretty good volume out of the heavy parts (anywhere from -15 to -13 RMS). Amazing what smoothing transients can do to volume without changing the overall mix.

 

[SouthSIDE Glen]

I absolutely agree that attacking the worst peaks manually before putting anything through the grinder is an excellent idea. What I find interesting, though, is how I tale a different approach to mastering the levels of the whole set. I'm not saying that one is right or the other is wrong or anything like that, just what works for me.

As far as leels go, I usually pick one of the quietest or most open mixes and master that first, using that as the lowest common denominator to master the rest of the songs to.

The way my ears see it, the quieter/more open arrangements (typically the sparser ballads) may have more "room" to be pushed before they completely fall apart, but they start to sound "pushed" to me long before they actually start to break apart. There's a ton of stuff in the CD bins that may not sound "broken", but they still sound very "pushed". I won't push that hard. I'll go until they just start to sound unnatural and then back off a dB from that.

Depending upon content, this usually will mean just a bit of a bump to the denser-content songs on the album to bring them to the same perceived volume as the fist song, plenty of room before being flattened.

I'd also comment that "same perceived volume" itself can mean different things to different folks. White Strat described it in an interesting vocal-centric way that I admit I never really looked at it - at least not consciously. I'd also add that same perceived volume from song to song is not always necessarily exactly what I want. Sometimes two songs can be on such extreme ends of the density scale that they need to keep a volume difference between them; if you pump the quiet one too much it sounds pushed, but if you throttle back the dense one it sounds kind of anemic.

It's then when the idea of song order comes into play; do I want these two songs of different perceived loudness next to each other? Sometimes I won't want the discontinuity; other times I may want to shock the listener with it. However the pieces seem to fit the best. But usually in any album that has some obvious swings between anthem and ballad in it, there will be at least some small modulation in perceived track-to-track volume over the length of the album for me. It may just be a dB or two, possibly not even noticeable when in pure entertainment listening mode.

Which is probably a good thing - mastering is like compression or reverb; unless you're going for special effect, if the casual listener can hear it, you probably have too much of it.

G.

 

[WhiteStrat]

Sometimes two songs can be on such extreme ends of the density scale that they need to keep a volume difference between them; if you pump the quiet one too much it sounds pushed, but if you throttle back the dense one it sounds kind of anemic.

I probably didn't say it well--but this is what I was saying as well. I don't want the quiet tunes to sound big relative to the big ones, nor the big ones to sound quiet relative to the quiet ones.

So they don't all come up to the same level. The big ones come up the most, representing the "noisey" peaks of the album, while the small ones represent the valley. (Considering how much headroom the quieter mixes have, they probably come up as much, if not more, than the loud mixes, but in the end they still sound quieter.)

The result (if I do it right) is an album that has its own dynamic ebb and flow--like a big long song of its own.

 

[miroslav]

Most of the ones I have on my album are failry dense mixes...maybe not quite "Wall OF Sound", but close.

I found that even with them, if the limiting pushed a bit too much, they sounded boxy...maybe even a touch dull...but LOUD!

 

[gecko zzed]

White Strat described it in an interesting vocal-centric way that I admit I never really looked at it - at least not consciously.

I tend to use a similar method to White Strat in that I use (in part) the vocal level as a way of judging consistency between tracks. When trying to balance different intensities (inherently loud songs versus inherently soft songs) I go for the 'what if I was at a concert?' approach . . . figuring out whether soft is soft because it should be, or whether adjustment is needed to achieve something that sounds more credible. If I have mixed the songs pretty well, it seems that if the vocals are okay across tracks, the music looks after itself. The other reference I use at times (and often in conjunction with vocals) is the snare, again because generally its level is about the same across different tracks (with roughly the same caveats as vocals).

 

[WhiteStrat]

I tend to use a similar method to White Strat in that I use (in part) the vocal level as a way of judging consistency between tracks. When trying to balance different intensities (inherently loud songs versus inherently soft songs) I go for the 'what if I was at a concert?' approach . . . figuring out whether soft is soft because it should be, or whether adjustment is needed to achieve something that sounds more credible. If I have mixed the songs pretty well, it seems that if the vocals are okay across tracks, the music looks after itself. The other reference I use at times (and often in conjunction with vocals) is the snare, again because generally its level is about the same across different tracks (with roughly the same caveats as vocals).

I follow the "in concert" train of thought as well! Everyone's rocking out for a couple numbers, then the band takes a break and leaves the singer at the piano. If I push that piano tune too much, it will sound like the singer suddenly left the stage and is now sitting on my lap!

Good call with the snare too. That's a great benchmark as well. (Unfortunately in this project, I've only got snare in about half the tunes.)

 

[miroslav]

I find that when going from loud, up-tempo stuff to more subdued, sparse cuts...if you add a second or two extra between the cuts...it helps "clear the palate" for the next tune and the difference between the two cuts are not as noticeable.

 

[SouthSIDE Glen]

If I push that piano tune too much, it will sound like the singer suddenly left the stage and is now sitting on my lap!

Well, if it's Diana Krall on my lap, I wouldn't complain too much . (Sorry, I had to swing at that softball. )

Seriously though, yeah, Strat, I get what you're saying a bit better now. Not that I disagreed with you before, I didn't; but now I find we are more in line than I thought. For example:

The result (if I do it right) is an album that has its own dynamic ebb and flow--like a big long song of its own.

This, IMHO (FWTW), is not only a very important point, but is a major part of the essence of quality mastering.

This idea is lost a lot these days when "mastering" often involves just making singles to throw up on the Internet (not that there's anything wrong with that - more on that in a minute), but when putting together anything from an EP to a full album, I think it's important for the producer and the mastering engineer to remember that are creating an album as a single entity, not just as a collection of individual songs. This plays not only into a certain level of "cohesion" of timbre throughout the songs without taking the individual character out of each song, but also into modulating the song order and levels through the course of the album.

To go back to the mastering singles trend of the day, a lot of people argue that the album is dead as an item that people listen to as more than one song at a time, and has been replaced by randomized or custom-ordered playlists. While this may be rather true, I don't personally see this as a necessarily permanent trend, or even if it is, as one that should affect how one masters an album as an entity, nor how they should master their singles.

G.

 

[leddy]

I'm surprised the RIAA does not push the industry to standardize on a certain loudness level. If it was set low enough, it would sure make all records sound better. Not that people couldn't ignore it and still do what they want - but then those who ignored it would run the risk of having their stuff sound like poop when it gets played on the radio and smashed by the broadcast limiters.

Doesn't the movie industy have a loudness standard? I think they do.

 

[WhiteStrat]

Well, if it's Diana Krall on my lap, I wouldn't complain too much . (Sorry, I had to swing at that softball. )

Seriously though, yeah, Strat, I get what you're saying a bit better now. Not that I disagreed with you before, I didn't; but now I find we are more in line than I thought. For example:This, IMHO (FWTW), is not only a very important point, but is a major part of the essence of quality mastering.

This idea is lost a lot these days when "mastering" often involves just making singles to throw up on the Internet (not that there's anything wrong with that - more on that in a minute), but when putting together anything from an EP to a full album, I think it's important for the producer and the mastering engineer to remember that are creating an album as a single entity, not just as a collection of individual songs. This plays not only into a certain level of "cohesion" of timbre throughout the songs without taking the individual character out of each song, but also into modulating the song order and levels through the course of the album.

To go back to the mastering singles trend of the day, a lot of people argue that the album is dead as an item that people listen to as more than one song at a time, and has been replaced by randomized or custom-ordered playlists. While this may be rather true, I don't personally see this as a necessarily permanent trend, or even if it is, as one that should affect how one masters an album as an entity, nor how they should master their singles.

G.

Yep. I guess I've sort of got a foot in both camps--old skool and new skool. I'm only moderately bothered by the loudness wars. I enjoy that last bit of punch I get out of my ad hoc mastering process. (The majority of my own music is electric guitar instrumentals, so it works for me there.)

But at the same time, I can't pursue that loudness at the expense of the album. I just can't see it as 10 or 12 separate entities, each to be treated in a vacuum. To me, the album is one long song. I write that way. I record that way. In fact I'll deliberately write and record pieces to balance and contrast with others. And if something doesn't fit--it doesn't go.

So the last stage--the mastering--absolutely has to serve that purpose as well. It has to further the cause of creating one piece of music (albeit with 10 or 12 "movements").