Trying To Get Things Right For ACX Submission

Hello all!

First I want to thank you for your responses and time here. I have found this forum to be a wealth of knowledge.

I am getting ready to narrate an audiobook for ACX for a friend of mine who is an author. I followed Koz’s guidelines for clip submission and have attached a recording. I greatly welcome any advice you can give me on how my home audio recording sounds. This is completely raw, as required, and I am trying to dig through all the information on how to use Audacity to edit and prepare my recordings for ACX submission.

Thanks again!!!

It fails, but it fails in a very gentle way. I’m going to see what little I can do to it and post the results.

flynwill wrote an auto analysis tool specifically for the ACX-bound. You don’t have to do multiple passes through “contrast” to see what you’re doing. It is good to be aware of what the specifications are to make sure you’re firmly in compliance and not “failure adjacent.”

Attached acx-check.ny. Copy that to your Audacity plugin folder and restart. Analyze > ACX Check.

Results attached.

Koz

acx-check.ny (5.6 KB)

RMS is intensity, density or loudness. This measurement gets worse if you like to pause between words or sentences for dramatic effect. Even compensating for that, the punch in the voice isn’t enough. So now it’s decide which processor to use to cause the least damage. The good news is the noise level should be low enough not to worry about it.

The killer performers have marginal sound files where you fix one thing and break something else.

Koz

Thanks Koz! You don’t think this is a recording level issue for the RMS? I have the gain on the mic on the halfway position. I have the Audacity recording setting to .5. I am aware of the ACX requirements, but being the complete noob that I am, I am flying blind on how to use Audacity to edit the product to make it work. High pass, low pass, normalize, equalize, noise reduction… it is all mind boggling. Do you have any recommendations or template to follow?

The processing is a little bit of a violin. It’s not “push that button and everything will be OK.”

I got the clip to pass (attached). Listen to the patch and then to yours. It’s not just volume. It’s more forceful.

I’m going through the patches and see which is the simplest one that will make everything OK.

Koz

Ah… hard to understand in words, but I definitely can hear the difference between the two samples.

How I got there. You should do this with strong coffee or tea. Instead of presenting a laundry list of settings, I attached screen captures of everything I did. You’ll still get lost, but it will be much prettier. Click on the graphics to see them larger.

Normalize is used more than once for different reasons.

You should install the attached limiter (limiter.ny). It’s a custom tool.

There is a custom equalization tool as well: LF_rolloff_for_speech.xml This is a little rougher to install.

Adding Equalizer Curves
– Select something on the timeline.
– Effect > Equalization > Save/Manage Curves > Import
– Point at LF_rolloff_for_speech.xml > OK.
– LF Rolloff for speech now appears in the equalization curve list.



Open the clip.

Select the whole clip by clicking just above MUTE.

Effect > Equalization > LF rolloff for speech (attached pix) > OK
Effect > Amplify > OK (default settings)

Effect > Limiter (attached pix)
Effect > Normalize (attached pix)
Effect > Compressor (attached pix)
Effect > Normalize (attached pix)

And that’s all there is to it [dusting off hands]. Let me know where you get lost.

If I did it right, almost anything you do with about the same microphone and environment should pass with those filters—and even sound like you.

Koz




LF_rolloff_for_speech.xml (299 Bytes)
limiter.ny (1.52 KB)

If you’re having trouble sleeping I can tell you what each step does and why it’s needed.
Koz

Wow Koz, thanks! I was able to follow it no problem. This brief audio sample didn’t include much fluctuation in my voice as might happen in an action scene. Will your suggestions cover if my voice gets louder so I don’t go beyond ACX’s requirements? Also, what did each step you showed me just do? (in layman terms)

You are required not to max out the sound channel by getting too loud. Make the sound meters go up to about -6 while you’re recording. Yes, you need to watch those and present at the same time. This kills people who have to put the computer in the next room because of fan noises. One poster attached a simple external monitor to the computer outside her room so she could see what’s going on. It worked perfectly.

That’s an actual picture from one of my actual voice recordings.

Successful presenters “get loud” by increasing the stress in their voice without actually getting louder. Another trick is back away from the microphone and then get louder. You’ll get the hang of that about the end of this audiobook or the start of the next one.

And yes, if you can keep the sound meters out of the red, this recipe should keep you in compliance.

LF_rolloff is a custom filter designed by Steve to remove all the rumble, thunder, earthquake and other low pitched noises from your voice, without, for the most part, interfering with your voice quality. Some microphones naturally create sound junk down there and it can screw up some of the other tools.

Amplify pushes the volume of your performance up as high as it can possibly go without distortion to prepare it for…

Limiter which suppresses only the loud peaks and leaves everything else alone.

Normalize (in this case) is used to prepare the sound work for…

Compressor which squeezes the loud and soft parts of the performance closer together in volume. It’s subtly different from what the limiter does.

Normalize (again) makes up for compressor volume errors and prepares the sound for the ACX peak sound specification. (no higher than -3.0). Normalize is set for -3.2 (slightly lower). If you set this too close to -3.0, the conversion to MP3 may violate the conformance.


Asleep yet? RMS stands for Root Mean Square which is a measurement of the DC Equivalent energy of a complex waveform using area-under-the-curve approxi…zzzzzzzzzzzzz

Koz

Thanks Koz that helps a lot. I really appreciate it!

Koz,

I have my mic in a box that is padded with foam. For making the recording you helped me with I was about 6-8 inches from the mic. The hard part is finding a good place to put the material I am reading from. I’m afraid that by putting it off to one side I will turn my head too much and it will affect the recording. Conversely trying to set it on top on the box the mic is in, gets it too close to my face and makes it hard to read. Any suggestions?

You might notice (unless you’re cutting some serious z’s right now) that the live recording specification has peaks at -6dB but ACX peak specification is -3dB.

Different goals. Your goal is good live volume without overloading and clipping (generally fatal).

Their goal is to make the work as loud as possible while including a little elbow room and more importantly, consistent from chapter to chapter.

That means even under perfect studio conditions, you will always have to make up that tiny difference.

Koz

Conversely trying to set it on top on the box the mic is in, gets it too close to my face and makes it hard to read.

Everybody sooner or later reinvents the sound studio (attached 1).

“If only there were a room where I could spread out my script and didn’t have to worry about the placement of the microphone…”

But all seriousness aside, some people do read from iPods, cellphones or other Personal Display Devices. I couldn’t do it, but I’ve seen it done.

Myles on the left there (attached 2) is either getting his lyrics from the phone, or he’s playing Angry Birds during the performance.

Koz

Screen Shot 2015-06-11 at 19.38.50.png852×476 430 KB

BiDirectionalRadioDrama.png729×567 540 KB

He’s the tenor/soprano.

https://www.youtube.com/watch?v=cUhdL_hUi3w

Koz

Nice video. I am reading from a Kindle in my clip. Baby steps I suppose. With enough duct tape I can probably get something to work!

Hi kozikowski,

I viewed your original post on in this string, back on Aug 02, 2007 (https://forum.audacityteam.org/t/trying-to-get-things-right-for-acx-submission/38743/1).

I’ve recorded all of my book chapters, and want to prep them ready for upload to Audible. I’ve taken some general tutorials to find out how to edit in Audacity - but some of the ACX settings don’t make sense to me. Are you please able to help me?

e.g. ACX requirement “Each file must measure between -23dB and -18dB RMS”. Yet I can’t find any function in Audacity that will allow me to set the limits. Also, how does this requirement function alongside the requirement “Each file must have peak values no higher than -3dB” (If it’s got to be between -23 and -18. how can it be max -3dB?) - I’m sorry: I’m not a tech geek at all, so all of this language phases me! I don’t even know what a “RMS” is.

In your original post, you attached a “Limiter” App which I downloaded, in case that could help me. However, I don’t know what program to use to open it (I tried Audacity, just assuming that must be the one to use. I’ve got Audacity v. 3.1.3 and it doesn’t recognise the “Limiter” file from 2007).

I’m sorry to bother you with all this - but any help would be very much appreciated.

Thanks.

I just read another of your posts, which talked about the Nyquest plug-in installer, so I’ve now installed the “Limiter” download in Audacity - one problem solved! However, I have no idea where to go from here, to get all of the requirements correct for ACX.

Any help from anyone would be much appreciated (but assume that I’m a total novice, and won’t understand any jargon - or how to use any of the Audacity Effects. Pointing me to the correct “effect” screen will be part one; helping me understand how to use the setting on it to get the results I need will be something totally different . . . !

Very many thanks.

This message thread starts in the dim distant past when ACX helped you out a lot more and Audacity a lot less.

ACX will no longer accept a sample submission for full analysis and comment like they did when I posted one. Audiobook Submission became insanely popular and everybody called in sick, so now you have to show up for publication correct and complete in one shot, without baby steps.

We offer ACX-Check which will tell you RMS (loudness), Peak (overload), and Noise (what’s there when you stop talking). Those are the three technical tests. There are two versions of ACX-Check and if you have an older one, it may not be able to analyze longer chapters. The new one is highly recommended.

https://wiki.audacityteam.org/wiki/Nyquist_Analyze_Plug-ins#ACX_Check

This analysis can be done with no human interaction, so ACX actually offers an on-line version of this called ACX Audiolab.

https://www.acx.com/audiolab

Obviously, they expect you to read well, be near-sighted, have a blue-jeans jacket, and no hair.

Audacity ACX Check will tell you all three technical specifications. ACX Audiolab won’t tell you background noise. Neither one will give you a theatrical test. That’s the one I failed. That’s the one you need a human for evaluation. That’s why we keep telling people to submit a short voice test to the forum. Without that, the first time you find out you can’t read would be when you submit all your finished chapters. No, that’s not desirable, but that’s the way it is now.

Working backwards: We publish Audiobook Mastering. That’s three tools, included in Audcity 3.1.3 (the current Audacity release), that will clean up any performance to automatically pass two of the three technical tests.

https://wiki.audacityteam.org/wiki/Audiobook_Mastering

This is the short version.

This is what the new ACX-Check looks like when you pass.

There is no shortage of studio problems when you try to read from home. Homes are noisy and the first time you find out just how noisy is when you try to read for Audiobooks. It’s perfectly normal to pass RMS and Peak and fail Noise. That’s another place the forum can be handy. We can sometimes recognize noises and give you hints and tricks to get rid of them.

And there’s a trick to reading a script and speaking into the microphone at the same time. Don’t perform head-on. You can use Oblique Placement (B) where your script is straight in front but the microphone isn’t.

That placement can partially make up for a “quiet” home microphone.

If you’re set up for announcing, post a ten-second voice sample to the forum and we can start there.

https://www.kozco.com/tech/audacity/TestClip/Record_A_Clip.html

Koz

assume that I’m a total novice, and won’t understand any jargon

A word on dB. Audio volume is measured in dB.

Koz