Overused Noise Reduction (I think)

Hi all – I have Windows 10 on my laptop and am using Audacity 2.2.2
First off, I just recently began exploring the world of audiobook narration. Unfortunately, my talent lies only in reading the manuscript. I evidently don’t have a technical bone in my body. I just had a completed project rejected by ACX because the “submitted files do not meet our noise floor requirement.” A max -60dB

This summer I completed two projects, which were both approved by ACX. The only difference I can think of in these files is that I used (and used and used :confused: ) the Noise Reduction feature on this latest one, producing poor sound quality. I’m not entirely sure why I did this. Rookie mistake, I guess. The problem is, all of the files I saved for this newest, rejected project are saved with the Noise Reduction effect. No raw files. My question is, is there any way to undo or fix the mistake without having to re-record?

(And when you respond, can you pretend you’re responding to a kindergartener? That would help me understand better.)

Thank you!

Usually over-using noise reduction will give you the required noise floor, but can get artifacts (damaged audio). And I think your files will be rejected if the background is unnaturally low (but that’s not what they said).

Does it actually sound bad?

Run the [u]ACX Check plug-in[/u] to see what you get.

My question is, is there any way to undo or fix the mistake without having to re-record?

That’s hard to say. Did you save your original un-processed files?

It’s good to know of a person is associated with your rejection notice. ACX acceptance goes (so far) in two steps. The Robot which works very much like our own ACX-Check tool, followed by Human Quality Control. The robot is fairly simple. It checks the three main characteristics of audio production: is it loud enough, is it noisy and is it distorted? That’s RMS, Noise Floor and Peak. This is how ACX Check displays that.

You don’t need ACX Check, either. That’s just a shortcut. You can do it with regular Audacity tools if you’re into pain.


Some of the tools give you answers as a by-product of doing something else. It’s pretty awkward. ACX Check was a welcome addition.

Human Quality Control is where you go to die if you passed the robot by beating your voice with a stick so it sounds like a bad cellphone. So you do have to create good quality in real life. You can’t make it up later in post production. I produced two April Fool spoofs where I made fake software that would turn badly recorded audio garbage into a polished studio production. It was pretty popular.

is there any way to undo or fix the mistake without having to re-record?

Short answer is no. We can’t remove effects, filters and corrections. That’s the bumper-sticker version. We can take out simple volume changes, but even that can have problems.

So that’s two errors. Now you know why it’s highly recommended to Export original (raw) readings as WAV protection copies.

I can predict the third. Put valuable work in two different places. Where is your work going to be if the dog eats your laptop?

I have an extreme version of that. I have an older desktop machine with valuable and complex layout graphics. This machine has two different hard drives. C: and D:. I copied finals to both drives. C: went into the bin and took the machine with it. I pulled out the D: drive, mounted it in another machine and just kept right on working. Normal people might copy work to thumb drives, external hard drives and even Cloud drives.

Now, a side trip. We publish a simple process for audiobook mastering, but it’s not the only one out there. Many people start the process with Noise Reduction. We don’t. Is that where you got the idea you needed that? It’s good to trace errors back and see if they can be avoided—and warn others.


Missed a step.

This is Audiobook Mastering version 4.

Most of that is a long explanation of three tools. I need to see if I can make it look a little simpler. It scares people.

pretend you’re responding to a kindergartener?

That’s a gimme, isn’t it?

You probably figured out loudness seems upside down. Maximum loudness is —0— and it gets quieter as the numbers get bigger. -3dB is the maximum loud an audiobook is allowed to get and -60dB is the maximum loud noise is allowed to get. They’re not simple numbers. -60dB is 1000 times quieter than —0—. It’s because your ear works weird.

The bouncing sound meters work in dB, but the blue waves on the timeline work in percent. 100% is 0dB. 50% is -6dB. The blue waves only display the most significant sounds. For example, you can only see blue waves down to about 3% or so before they turn into blue mush. That’s only -30dB out of 60.

You’re a celebrity unicorn. Nobody passes their first book and fails the second.


Here’s another one.



On my end, the files sound pretty good; HOWEVER…I did notice what seemed to be a lack of required room tone. Enter Noise Reduction overuse.

Unfortunately no. I was pretty confident in my completed work, so I overwrote my raw files with the edited version. Another rookie mistake and a lesson learned the hard way.

Thank you for your input!

The rejection email I got came from an actual person…both times (so embarrassing).

I’m not much for pain, including the pain I’m experiencing trying to get ACX Check into Audacity plugins.

Yes, lesson learned the hard way. I’ve already re-recorded the entire book.

I got the idea for Noise Reduction from a tutorial I watched on voice recording. But this was more for voice acting and submitting auditions for commercials, etc. Not for audiobook narration. I’m realizing there’s a difference. Thank you for the resource. I plan to check it out.

I just might put that on my ACX profile.

Sincerely, thank you for all of your input and for the resources. Grateful for your knowledge and willingness to share it.