"Solid" waveform bad? Effects workflow critical? (Prepping for ACX)

[JeffOYB]

I'm prepping narration files for an audiobook.

The files are now all edited for mis-speaking and for background sounds and clunks.

Now I'm trying to make them all consistent in volume so they meet the ACX requirements of RMS, Peak, etc.

I'm listening to the results on some earbuds that might be failing. Offhand the volume and quality sounds good. I'll try to get some better earbuds.

Some things I notice, however, are shocking me and getting me worried.

I've been using the Match Volume function to adjust my Total RMS and Peak Amplitude to fit the requirements. I notice how some of my source files display waveforms that have varying sizes and degrees of "fuzziness." I did use 3 different microphones in recording these files. My concern is that sometimes when I change the Total RMS then apply a Peak that the resulting waveform sometimes ends up looking like a SOLID BLOCK of sound -- like ALL the sound gets increased to a certain amount and then clipped back to fit the -3db Peak. The spectral waveform still looks the same though... And the sound still seems fine. ...Especially when listening through my (undamaged) laptop speaker. Also, the waveform seems to only be chopped on the top-side, not symmetrically limited. Weird!

I'm also wondering if these changes are destructive. Or can I keep adjusting the levels using various strategies until the result seems best?

Then there's workflow. If I start with a somewhat quieter file it might not have hardly any louder spikes so if I first apply the -3db Peak it maybe doesn't change much, then if I apply a change after that to Total RMS I might then end up with a lot of peaking and need to apply the Peak again... 

Thanks for any thoughts on this... Or helpful links...

Here are screenshots of before and after.

Before at Peak -.7dB and Total RMS -29dB:

After has Peak -3dB, Total RMS -21dB. ...Shows a rather "solid" waveform with an assymmetric top side:

Here's an After shot that started as an even quieter waveform and ended up REALLY solid looking, also asymmetric. It fits the specs but LOOKS really weird to me:

Here are some more Before and After's....

Here is an overly quiet file -- Peak -5db, Total RMS -31db:

Now I've increased the Total RMS to -17db:

Now I set the Peak to -3db -- and the Total RMS comes out at -20db. I note that this file didn't end up with as "solid" looking of a waveform as some of the others:

Now, here is a file where the waveform looks "normal" to me and it is within specs!

[SteveG_AudioMasters_]

There's quite a bit to be said about this...

Yes there is an order in which processing should be applied, and they will make a significant difference to the results you get. But in order to get consistent results, you have to record in a consistent way, and that means paying attention to a lot of details. So you have to keep your recording situation - primarily the mic, your positioning, and recording level - to be the same for all of the production. At this point, you'll have a recording with a relatively low mean level, and a lot of peaks. What you need to do is find the very highest peak out of all of these recordings and normalize all the tracks by the same amount, and that amount is when that highest peak is at 0dB. The reason for doing this is that all of the limiting/compression processes in Audition are normalized at that level, so that's where you need your highest peak to be. At this point, any limiting applied would be applied equally to all tracks. What you need to do at this stage is to limit those peaks so that the top 8-10dB of them is reduced in level, leaving the bulk of the audio at the same level. What you should end up with then is audio without massive peaks that sits at around -9 to -10dB peak level. Now you can normalize the result so that the highest peak is at -3dB. If it's a well-recorded file, you should end up with an average RMS value that fits the ACX gamut of -23 to -18dB.

How do you do that? Well, initially you can scan each file for its peak level, and when you've found the highest one, make a note of how far below 0dB it is, and how much you have to amplify the file to get it to 0dB (you can use Amplitude Statistics for this - the peak amplitude is at the top of the list, and it's the number nearest to 0dB you need to make a note of).

Like it says on the ACX mastering page, you really shouldn't need to be doing anything else, and especially anything involving compression - limiting only is what's required. No messing about with total RMS or anything like that is required at all - it will almost inevitably make everything much worse, because you won't know where any correct levels really are. Once you've got the levels right, only then should you think about anything like NR if there's anything consistent that you can get rid of.

All this is approximately what it says to do on the ACX website - there is no processing regime that has to be absolutely rigidly followed  - except inasmuch as you have to be consistent about the way you treat everything. Mostly this is about what you shouldn't do, not what you should. The relevant page on the ACX site is here: Audiobook production pt 3 

[JeffOYB]

THANKS!

You'd mentioned looking at ACX requirements before when I asked how to do something -- and I saw all of the requirements but no how-to info. So I was completely baffled.

I never knew til you posted the link above that there was impt Mastering how-to info (beyond the specs) in the *blog part* of the ACX site! This is now wonderfully helpful!

I still wonder why the waveforms are assymmetric in terms of top and bottom...

[JeffOYB]

PS: I note that I could understand the ACX mastering info. And I am greatly appreciative of the link. But I have to say that I understood far less of what Steve wrote -- esp the area about the peaks being 0db and 8-10db -- but I did sense some overlap, which was encouraging. I'm not sure why the difference, but it might be because of jargon. Since I'm a layman I really don't know what I'm seeing or how to describe it. Nor do I understand technical concepts. But I do know how to enter numbers into effects. Somehow the ACX mastering blog was able to get through to me.

[JeffOYB]

Wow. What a ride! Well... I had already done what ACX (and common sense) suggests for mastering, which is keep backups at each stage. I read the mastering article in the ACX blog and re-did a file that had a particularly solid-mass waveform. The ACX approach leans hard on the Hard Limiter and the resulting file met the specs and didn't look as solid. But when I listen to both files I cannot discern that one is better than the other! The solid one was a bit louder. And maybe different in other ways, too. But I can't tell if it's much better or worse. My ear is un-educated. Is it possible that there are several ways to climb the mountain? (I have watched many Audition tutorials explaining that various functions can be used to get the same or similar result.) I've bravely uploaded a couple short samples of my audio so anyone can see what they think about them! Hack maybe both samples are unacceptable and would be rejected by ACX  due to quality rather than specs. I suppose I should just submit them and let them decide. But maybe Steve will lend an ear?  🙂  OK, here are links to two 30-sec clips...

Sample remastered using ACX tips (Hard Limiting): https://drive.google.com/file/d/1dPN4zHuJLWRSRsMKtyn4kweCcF1wgOA_/view?usp=sharing

Sample mastered using YouTube tips (Match Volume) for ACX prep:

https://drive.google.com/file/d/1kM-Q4Kb0h0IspfolCKSSfQK5GNgRaHMc/view?usp=sharing