Does this pass ACX standards?

Hello all! This is my first post on here so I apologize if I’m doing it wrong. I’m looking for any feedback… whether it be technical, performance-wise, etc. I have profound hearing loss in both ears which makes the opinion of others invaluable! I have ACX check on my Audacity 3.2.3 on Mac High Sierra 10.13.6. and it passes on there, however I read that sometimes even that doesn’t matter and it still won’t pass the real deal. Again I would love any and all feedback… :slight_smile: I’m trying to do this seriously for a potential gig on ACX.

I’ll attach some smaller samples but if anyone would like to listen some more for a better idea I will post a link to my ACX page for more samples.https://www.acx.com/narrator?p=A2KVUZ8XM55MHO

Thank you!

I read that sometimes even that doesn’t matter and it still won’t pass the real deal

ACX Check is only testing volume, overload damage, and background noise. Those are the basics. You can have other problems. You can read too slow, you can have reading-in-a-kitchen room echoes, you can stutter.

Your book must be for sale on Amazon. Don’t read anything on this forbidden list (scroll down).

https://www.acx.com/help/200878270

Don’t read a cookbook.

As a fuzzy generality, your book should have Plot, Characters, and Setting. There are variations to this. One of my favorite presentations is by Joseph Campbell who discusses “The Hero’s Journey.” In that case, he’s talking about Plot, Characters, and Setting

You can carry regional accents too far and make it so a casual listener may have to roll the story back and listen carefully to understand what you said. But regional accents are not a death sentence. Just don’t go nuts. I used to be sent out on large, group-effort television productions because I could speak both New York City and Atlanta.

The plural of y’all is “all y’all”.

When I tried this audiobook thing, they had a sample review where I could submit a short segment for evaluation. My submitted file was technically perfect, but they hated my mouth noises. Now, you have to submit the completed book. Do they still do that? We wish they hadn’t done that.

I didn’t go into the longer presentation, but I can listen to a story in either of those posted voices.

How did you get the swing, meter, and emphasis if you can’t hear what you’re doing?

Koz

Thank you so much for your response Koz! Thank should clarify - I have profound hearing loss in both ears but do have hearing aids. Even so, I cannot hear subtle sounds that well like mouth noises for instance. This is why I depend on others a lot for opinions on sound and speech as I’m slightly “tone-deaf” so to speak.

Thank you for your feedback!

Would you say technically they are sound? Any recommendations on what I can be doing better? Appreciate the time you took in responding sir!

I think they’re fine as they are. I think you are more likely to run into oddball problems like needing to have your book for sale on Amazon before they will consider it.

https://www.acx.com/help/acx-audio-submission-requirements/201456300

All your chapters have to match, etc. It is normally recommended that you wear sealed headphones and listen to yourself in real time.

But that’s not going to work for you.

Koz

One odd note. Room Tone or background noise should not be dead quiet. Yours is good at -67dB or so. Anything from there into the -70s should be OK. They don’t like submitted noise in the -80s and -90s. That usually means the poster has applied limiters, filters, and Noise Gates to the work and that can damage the sound.

This problem just kills many home readers. They’re fighting natural house and room noises, plus low volume and noisy home microphones. Everybody reaches right for the Noise Gates and other post production tools.

If you can’t walk away from a simple recording with background noise in the -60dB range or slightly better, it’s going to be a long day.

Koz

Despite what ATX-Check plug-in says,
IMO there’s too much hiss-noise on that for a profe$$ional performance …


A hint of Noise Reduction plus hint of expansion can bring the noise-floor down by ~6dB, into the <-70db professional zone, without ACX noticing you’ve applied those forbidden tools … https ://audio.com/anonymous-audio/gambling-man-audacity-b-a

Your equalization is good, but IMO benefits from filling-in notches @ ~850Hz ~2800Hz …
gambler spectrum  B-A EQ.gif
https://manual.audacityteam.org/man/plot_spectrum.html

Thank you so much Koz for the input - I do what I can with my disability. I do have a headphones that don’t seal but they hook up to my bluecoil interface that I use. I use an AT2035 (I think that’s what it is) mic. It is USB unfortunately…

Trebor - Appreciate the reply. Unfortunately I do not understand much of it as I am what many would call a beginner. With that said, I do noise reduction on some tracks if needed where the noise floor is -55db or so. I do the number of the beast 6, 6, 6 whenever I apply it.

For expansion - this is a new concept and I do not understand it but am willing to learn about it.

So with these two tools - will it eliminate the ‘hissing’ noise you mentioned earlier that would hold me back from a professional sound?

I listened to the audio you uploaded Trebor with the side by side comparison. I assume that’s with the expansion and noise reduction applied in the second half?

Also how did you do the notch-fill in? And what exactly does that do?

Forgive me for the myriad of questions but you’ve supplied a great deal of info and I’m trying my best to digest it XD.

You guys have been a huge help. And I thank you kindly for the time you took.

Okay Trebor, after having some time after the kids went to bed I’ve looked at the expansion item you posted about. Clicked on it and read as well as I could (computer-language illiterate) and still failed to grasp what I need to do.

If I can get away with Audacity’s noise reduction then perhaps that’s what I’ll just do.

Also I tried looking into the equalization notch-filler you mentioned, again, without success.

Seems I need to hire someone to help me set up properly!

My suggestion would be use noise-floor of -70dB, rather than the bare-minimum -60dB specified by ACX.
That would probably mean employing some Noise Reduction on all your output, (the least necessary to get you close to -70dB).

It’s similar to a noise-gate. It pushes down the noise-floor, but only acts when you’re not speaking.
Must avoid setting an expander/gate to squelch the noise-floor to minus-infinity, as ACX don’t like that.
I believe you can get away with using an expander to attenuate the noise between words by 6dB without listeners & ACX noticing you’ve used processing to do that.

With Noise reduction & expander effects IMO your noise floor can be -70dB without the audio sounding processed, (i.e. the effects have been applied transparently).

I used Audacity’s filter curve to push out the notches ,
(but there is a free AutoEQ plugin which will automagically remove such dents).

Removing the notches in the spectrum removes your room from the recording.
(the physics of the room added the notches via destructive interference).

Thank you, that is all very good info! Just two more follow up questions if you would be so kind.

What do you use as an expander, I’m not finding any good plugins or tools for that.

What settings did you put into the Filter Curve to push out my notches?

Thank you so much!

6, 6, 6 is the starting point for noise reduction. It usually gets Home Readers into the -65dB to -70dB noise range. It’s gentle, clean, and nobody can tell what you did. If you need more moxie, you can go further up to 9, 6, 6. 12, 6, 6 is the limit. You (we) can start to hear artifacts and damage beyond that. If you can’t reach specification with 12, 6, 6, you got more serious troubles than plain backbround noise.

There is one condition where Noise Reduction fails. There is a microphone connection error where harsh data signals leak into the voice. I’ve been calling it Yeti Curse, from the popular USB microphone. The whining mosquito sound appeared so often, we designed a filter specifically for it. Mosquito Killer. The reason Noise Reduction fails is the pitch of the whine. It’s not Gentle Spring Rain In The Trees (fffffffffff). It’s “Baby Screaming On A Jet.” You can’t ignore it.

Even worse for the Home User, the only sure way to solve screaming mosquitoes is change the computer. It’s a hardware error.

If it makes you feel any better, ACX has provision to measure your file specifications.

https://www.acx.com/audiolab

See how your audio measures up to our ACX’s submission requirements on 8 important metrics including RMS, peak levels, bitrate, spacing, and more.

ACX-Test and AudioLab are cousins to each other. However, unless they changed it, they don’t measure noise.

Koz

The free version of the Couture plugin …


At one time it was possible to share Audacity EQ curves presets, but not now :frowning: ,
we have to redraw them, in this case with millimetre accuracy.

Okay Trebor, I’ve used the Couture program you’ve recommended using settings similar to what you recommended in a different post that I found on here.

I have added the peaks in the filter curve per your drawing (got it as close as I could!) I also have the low roll off for speech factory preset in there as well. So my macro is running your two custom peaks and the the low roll off for speech for filter curve. (see attachment)

I also use a de-clicker plugin and I was wondering, is it a good idea to run Couture and de-clicker BEFORE my macro or after my macro? Does it really make a difference either way?

I’m including a few screenshots from my research these last few days.

Thank you again for corresponding with me and helping me with this. My numbers look amazing after these settings and I am going to upload a renewed sample that has used these settings per your recommendation.

Please let me know if you hear an improvement.
Screen Shot 2023-01-31 at 8.32.09 PM.png
Screen Shot 2023-01-31 at 8.32.30 PM.png
Screen Shot 2023-01-31 at 8.33.00 PM.png

Also - Koz, yes I’ll be sure to check out the metrics on ACX that is a great tool I should think.

I feel like I’ve been having good luck with the 6,6,6 noise reduction this far…

However, if I have success using Couture, De-clicker then perhaps I won’t need the noise reduction anyhow! If I’m thinking right that is…

One production note. If you make a mistake in reading, don’t try to correct it later. Stop. Note where you are. Go back immediately to the next nearest sentence break and read the whole group again. Trying to correct a one-word error the next day will drive you nuts and will probably fail.

ACX puts great stress on having your chapters and reading techniques match. They hate, as they put it, “distractions.” Having your pitch or emphasis change in the middle of a sentence is distracting.

Once you start reading with your chosen Effect, Filter, and Correction Set, that’s it until the end of the book. No more changes. There is no getting a New and Improved Microphone in the middle of a book or updating your studio. You should also avoid “hand tuning” chapters or sentences. You will not be able to hear what you’re doing.

It can also be said that the fewer Effects, Filters, and Corrections the better.

Koz

You’ve drawn the filter correctly …
filter is spot-on.gif
but it’s not been applied: if you look at the spectrum it still has the notches …
Gambling Man 2 has not had EQ corrected.gif
Looks like Couture has been applied to “Gambling Man 2”, & in a transparent way:
If I hadn’t heard the original, which has audible hiss, I would not be able to tell an expander had been used.

If you’re using a de-clicker, I would use couture after it in the effects chain,
as any clicks in the “silence” between words & sentences could cause couture to let more noise through.

I would not leave out Noise Reduction in this case,
if anything I’d be tempted to use a fraction more then 6,6,6.

That is interesting as I’ve edited the first option in my macro ‘filter curve EQ’ according to your drawing (low roll off is still there, is that ok?)
Ok then shall the order be 1. De-clicker, 2. Couture, 3. Macro, and 4. Noise Reduction (every time)? More than 6,6,6… are we saying perhaps 9,6,6 is better?

I took the wav 2 file I just sent you completed last time and applied the EQ curver filter (just the two custom peaks, no roll off) and have pictured here below.

I wonder how come my macro didn’t apply it… perhaps the low roll off is buggering the process… hmmmmmmm. :confused:
Screen Shot 2023-02-01 at 6.23.03 AM.png
Screen Shot 2023-02-01 at 6.22.23 AM.png
Screen Shot 2023-02-01 at 6.18.32 AM.png
Screen Shot 2023-02-01 at 6.18.12 AM.png

That’s useful feedback, thanks.

The default setting for Noise Reduction is due to be changed to 6, 6, 6 for the upcoming 3.3.0 release.

Peter.

ACX Mastering Macro is a series of tightly integrated tools. The first one, Filter Curve is the Rumble Filter. It gets rid of sub-sonic, barely audible thump, rumble, earthquake, and thunder sounds. It also gets rid of very low pitch errors that many home microphones have because (1) rumble is expensive to suppress—home microphone buyers are acutely sensitive to price, and (2) most people can’t hear it anyway.

It pre-conditions the voice for Loudness Normalization which, as it says on the tin, boosts the overall volume of the work to satisfy the requirement of ACX RMS (loudness). Rumble in the voice can screw this up. Rumble filter is a cousin to the sound filter that many field, movie, and outside broadcast engineers use to suppress wind noise.

It’s not unusual for a loudness boost to cause the tips and peaks of the work to overload or exceed 0dB, so the process is followed by soft, gentle peak limiting. You would think peak limiting would affect RMS (Loudness) and it does, but not enough to throw the work out of specification. You might also think that any work that exceeds 0dB would be permanently damaged, but no. That’s why Audacity works in 32-bit Floating format. While you’re editing, Audacity doesn’t produce overload damage.

Peak Limiter has a goal of -3.5 dB instead of -3dB to make up for possible minor errors in converting the voice to MP3 for submission.

Not bad for three simple tools or one quick Macro.

If you do a reasonable job of studio construction and performance recording, it’s possible to read a chapter, apply the Macro, and you’re done. Export a WAV Edit Master, and only then the MP3 Submission File and go on to the next chapter.

Maybe in some conditions you may need Noise Reduction of the Beast, but that’s it.

Oh, one more. If it’s not needed, Mastering Macro doesn’t do anything. It has no affect if you apply it twice. You might notice a slightly stiffer Low Rolloff, but nothing else happens.

If you have theatrical and sound quality changes, you should probably do them all before the final Mastering Macro.

Koz

There is a housekeeping note. When you finish reading a chapter, export a WAV protection copy right then. Poor levels, errors, fluffs, and all. There’s nothing quite like having Audacity or your machine go into the dirt as you’re editing and taking your chapter with it.

Audacity Projects are not recommended for protection backups. Projects only open in Audacity and it’s not the worst idea to save progress Projects as you go. You don’t want anything to happen to that RAW protection WAV file.

If you need to use the backup, don’t edit that file. Edit a copy.

Koz