FINALLY! After a month of waiting for my submission to see if it was ok, this was the response I got. They said their turn around time for responses was 10 business days. It took them a month. I’m starting to believe that ACX isn’t the best way to go about partnering up to read audiobooks. I digress. Here’s the response from my audition for their standards.
Hello Maria,
I have reviewed your sample file.
This file is not ACX Compliant
Suggested improvements:
Recording: The fidelity of your recording is very good. It sounds like you have a quality microphone, and it’s set up properly. Your narration itself was well performed. Great job.
Editing: The flow of your title is very good. It sounds like your edits have maintained your narration’s natural pacing. I heard no extraneous noises that needed to be edited out. Very god work on your editing phase as well.
Mastering: The only real issue with your file is that the overall volume was low. Your RMS volume was -30dB. ACX RMS Requirements ask for an RMS of between -23 and -18dB. Since the volume of your file was close to being ACX compliant, only minimal dynamics processing should be necessary. If you apply a limiter to bring down the loudest peaks in volume, and then set the makeup gain (also known as output or output gain) to compensate, you should be right within spec.
I honestly don’t have the original file. But I can tell you all I did was do a basic recording, and used noise removal. I can try to record another bit and just upload again if needed. As
koz has stated before. The clarity in general, as well as the way I read should be good enough itself, and it looks like the RMS was the only issue( still have no idea how I’m going to do that ). Thanks again guys!
Ignore the foreground - background buisness, just select some audio and click on one of the “measure selection” buttons , the RMS value will appear in the white box next to the button.
I have the sad impression that the last 13 pages haven’t been of much good.
Well, you should now have installed some of the plug-ins that Koz proposed.
The limiter is here: Missing features - Audacity Support … er_.282.29
Normalize to 0 and measure the Rms value, either with “Contrast” or the Wavestats plug-in.
Calculate the difference from the measured value to the target levl of ~-20 dB (-18 to -23 dB) and add 3 dB for the head room.
Example:
-31 measured minus -20 minus -3 = 14 dB.
Run the limiter and enter -6 dB (as I do usually when the size is big) or enter -14 dB directly.
Repeat the steps (normalize, analyse, limit) until you’re close to -17 dB (-20 dB minus -3 dB).
If you have e.g. -19 dB, you can finally amplify by -3 dB to get some head room and the overall Rms level will be at -22 dB.
Ready for export…
I remember vaguely that I did much more with one of your files.
It seems that I’m more austere than the people from ACX.
I am catching up in this old thread. ACX requirements affect my work too but I have not had great difficulty with them.
Koz correctly describes the behavior of Noise Removal but is wrong to imply (with “in order to”) that this is by design. No, it is a defect. 2.1.0 promises a better effect, renamed Noise Reduction, which should be better at subtracting noise behind other sounds and less likely to damage the sound that remains.
I do use the present Noise Removal but no more than a slight 6dB reduction. Controlling noise at the source is best.
Back in the Paleolithic, we would judge noise by watching a sound meter bounce. In my case it was the Hewlett-Packard 334A Distortion Analyzer used in straight voltmeter configuration. A later Sound Technology meter worked in a similar way. The instrument had an RMS characteristic and everybody had the same meter. It didn’t work with popcorn or other extreme noises, but it didn’t matter because noise like that would kill your show automatically or cause a test to fail.
Since the Audacity sound meter characteristic is peak (or peak-to-peak since it reads both sides), it’s readings will reliably be higher or worse than the same sound passage measured any other way. There was a recent discussion of peak readings displayed in the fractions of dB, so I’m not making that up.
By extension, if my readings are much lower than a software measurement, there is something very wrong somewhere. I recall some software developers refer to this as a Rational Test.
One of the things that can go wrong in software is the selection method. If you select a portion of the show by accident, noise measurements will be trash. One recent poster stopped talking for his noise measurement (room tone), but in order not to “waste the time,” decided to shuffle papers, reorder his desk, and move a few pieces of furniture around.
Thanks to the above instructions, I was able to normalize the audio to the ACX requirements.
Now I would like to invite the help of experts to listen to my sample and tell me if the audio quality is good for audible? https://drive.google.com/open?id=1OO77DU6ZBk6y1UkmYDGWAW1zBG9k2BEy
The mic is Blue Yeti.
And is this a good place to ask such a question? If you know better places, please tell me.
Thanks.
If you must use a gate, it should be consistent, and not leave bits behind.
A gate with ~100ms (~0.1s) attack/decay it will remove the remaining bits of noise.
It’s 10 business days to post a perfect reading. If somebody has to sit, listen, analyze a bad reading and write a report, it’s going to take a lot longer.
The submission passes ACX, but they have a preference for mono over stereo.
Drop-down menu on the left > Split Stereo to Mono.
Delete one track.
They have a failure called Overprocessing. They might not like the “Blackness of Space” silences between words, particularly if you used the noise gate wrong.
