Indeed. Thanks, Paul. 
I’ve been following this amazing lady, Amy Walker (no relation as far as I know though I do have a cousin Amy back home in England, heh):
And here is a very interesting look at English in the Globe Theatre at the time of Shakespeare.
Cheers.
I can’t reproduce the static effect, so the best I can do is post the actual static clip. It’s a mono WAV file just under 10 seconds.
http://kozco.com/tech/audacity/clips/AMStatic5.wav
Loop it if you need more. I’ll continue to try and remember how I got there. There’s no trouble at all creating a mushy pseudo static by beating up Brown Noise, but I’m missing the one mystery step I used to create that honky snap typical of AM electrical interference. Fortunately, I saved the Project.
Koz
Where have you read about the ACX measurement habitude?
I thought here, possibly in the form of Steve’s repeating his conversations with ACX — buried somewhere under 20 chapters of comments.
I can even remember some of my comments, but not accurately enough for the search system to hit.
“If they’re measuring noise flat instead of A-Curve, then that leaves C-Curve, semi-flat and X-Curve no preconditioning at all.”
I worried specifically that anyone producing a just-made-it show by measuring in A-Curve is going to go straight into the mud when ACX measures it in C-Curve. In particular, pure Mains Hum isn’t measured in A because it’s barely audible.
Koz
Does this code help?
(defun brownian (gain) (scale-db (+ gain 17.2) (lp (noise) 40)))
;;
(sim; mix of soft noise...
(brownian -20)
; ... and sparkles
(scale-db -6 (quantize (brownian -6) 4)))
You can play with the gain values. The value “4” indicates the amount of spikes, so 10 would burry the signal under without adjusting the gain.
Here’s some rough and ready Nyquist code for a “radio effect”:
(apply to a reasonably short MONO track using the “Nyquist Prompt” effect)
;; Bad AM Radio
(defun bandpass (sig minhz maxhz)
(lowpass4
(highpass4 sig minhz)
maxhz))
(defun modnoise (dur)
(let ((dur (+ 20 dur)))
(clip
(extract-abs 20 dur
(stretch-abs dur
(sum
(mult (lowpass8 (noise) 0.1) 200
(fmosc 190 (mult 1000000 (lowpass8 (noise) 0.1))))
(mult (lowpass8 (noise) 0.1) 2000
(fmosc 180 (mult 40000 (lowpass8 (noise) 0.1))))
(mult (lowpass8 (noise) 0.1) 2000
(fmosc 198 (mult 1000000 (lowpass8 (noise) 0.1)))))))
0.05)))
(sum (mult 0.3 (modnoise (get-duration 1)))
(mult 2.5 (bandpass (clip s 0.2) 600 2500)))
Greetings,
Chapter one is uploading now, both raw and processed. It seemed to make no difference choosing the option to ignore the orphaned parts when the error came up at loading again.
I think I hear less of the ambient noise after processing this one.
I’ve also sent the foreword and chapter one 1 ACX to see what they make of them.
With my uneducated analysis, however, it seemed that chapter 1 hit 27 on the RMS so that may be a problem for them.
Cheers.
Mr. Grathwahl from ACX has released more information about how the company tests submissions.
We are applying A-Weighting to measure RMS, but we find Z-weighting to only measure ~3dB RMS louder than A-weighting in most cases. Indeed, we do A-weighting because transients have less of an impact on the final measurement. We analyze our files’ RMS levels in 60sec time windows, which may seem odd to many. We have both technical and business reasons for doing it this way, though, and I would want users to remember that their ears are always the best judge of a great production. If something doesn’t strictly meet our requirements, but it sounds great, we’re still going to take it without requiring any revisions or reviews.
Last week, ACX hosted a “narrator knowledge exchange” where we held seminars for editing and mastering concerns, attended by our users. Most of the Audacity users I instructed had been using the program for years but were entirely unaware of Chains. I think that’s all the indication we need that specialized tools are a solution that should be legitimately considered. Also mentioned during this instruction session was stats.ny – if you’ve used this plugin for Audacity before, or have heard of it, I’d be curious to know…
Is it essentially a grep’d version of SoX’s stats command output? If so, we’re using the same metrics to analyze our audio!
Mr. Grathwahl is not on the forum. Permission to use this information was by email.
Koz
For unweighted (Z-weighted / Zero Weighted) RMS, there is also RmsCalculate.ny https://forum.audacityteam.org/t/wave-stats/18863/6
This does not have the 30 second duration limit of the stats.ny plug-in.
Note that the A-Weighted measurement in stats.ny may not be accurate because it hasn’t been calibrated [ToDo:]
No it isn’t, but I’ll bet that the SoX stats command will be spot on accurate, so I can use that to calibrate the Audacity plug-in.
@MichloIW
How did your submissions go?
Koz
I haven’t heard back yet, mate. The instant I do, I’ll post here.
Cheers.
Curious.
SoX stats only shows unweighted rms, NOT A-weighted. http://sox.sourceforge.net/sox.html
I’ve just tested against the table published by CFFET (http://cffet.net/noise/noise_ch2a.shtml)
The “Wave Stats” plug-in A-Weighted measurements are accurate to within 1 dB for frequencies between 12.5 Hz and 16 kHz and with 1/2 dB between 20 Hz and 10 kHz.
It goes out a bit above 16 kHz, measuring 3.3 dB too high at 20 kHz, though that’s not really relevant to speech or music.
I am somewhat disappointed.
You should really have given the equalization a try. The audio sounds instantaneously more “pro”.
Just to recall the effect of equalization (on the raw-version):
The presets are here:
Cattle-curves.XML (27.5 KB)
The hum at 120 Hz is still very present on the processed first chapter
I think that you should be more aggressive when applying the noise removal, -12 dB is not enough.
There are some queer, louder parts in the track, are they punched in? For example “try to keep UP”.
Greetings,
I just got this:
"Hello Ian,
Thanks for contacting ACX!
I’ve made our QA team aware of your recently uploaded sample. They will retrieve the audio and listen to it and get back to you with notes. This can take up to 5-7 business days, so thank you in advance for your patience."
Cheers.
No, nothing is punched in. Some parts need to be louder as they are dialogue with people yelling. If you’re not meaning those then I don’t know, I didn’t notice them when I heard it back a few times.
And what do you mean I should have given the equalization a try. I DID! I used Steve’s roll-off for voices.
What exactly are you referring to please?
I’ve lost track of which is the most recent version 
(I hope you haven’t MichloIW 
)
I’m sorry, Robert, I know I don’t have the trained ear that you do but I hear no difference between your sample and my processed chapter 1.
I also don’t hear any hum.
I’m listening through AKG K240 headphones, BTW. I do appreciate your feedback but I’m not sure what to do with it.
Anybody else hear what I’m supposed to be noticing please?
Also, I added a few of my latest auditions to the feedback folder for something different.
Cheers.
I’m referring to Fledgling voice artist seeking counsel - #199 by Robert_J_H
Maybe you’ve just not seen this post.
In fact, the spectrum is uneven and thus the Rms value gets so low. We should actually use a multiband-compressor for this.
However, that’s unrealistic in Audacity and we’ll stick better to equalization and peak-limiting.
And the loud parts are:
From those, mainly “up”, “go” and “No” disturb the most in their quieter context.
Here’s my full version of the first chapter
https://www.dropbox.com/s/7o0ff4zjbrdrcyw/Cattle%20-%20Chapter%201%20(rjh-version.flac
I’ve not used a de-clicker.
Note that it sounds about 5 dB louder than your version (-21 dB-lin RMS )
I doubt that the ACX people use the A-Weighting, for the simple reason that bringing the value between -18 and -23 dB would mean excessive compression/limiting. No one wants that in an Audiobook.
Curious.
SoX stats only shows unweighted rms, NOT A-weighted. > http://sox.sourceforge.net/sox.html
This would not be the first time the specifications have been scrambled.
I’m going to formally invite Mr. Grathwahl to join us on the forum. I think it might be a boon to all of us.
Koz