I would like to begin to record a voice demo for web e-learning, however I am not sure if my noise level is low enough. Especially if I have to turn up the volume or use mastering techniques in the final processing this may increase the noise level even more. I am wondering if there is a typical output level or range that e-learning web audio falls into? Or does it solely depend on the client?

Thanks so much!

The only hard-and-fast standards I know of are the [u]ACX audiobook standards[/u]. There is an [u]ACX Check Plug-in[/u] for Audacity.

Those standards are tough so if you think you have a problem you probably wouldn’t meet them.

You can post a sample of your work on the forum. As a fuzzy rule, if you can pass ACX, you an do almost anything else. ACX technical standards are very close to broadcast They’re cousins.

Record and submit a 20 second test according to this formula.


We publish an Audiobook Mastering Suite of tools.


Technical standards will only take you half-way. When you actually do submit for publication, the work also goes past ACX Human Quality Control. That’s where you go to die if you record in a bathroom or you can’t read out loud. We don’t have any way to test for that past personal experience and opinion.

Most home readers fail noise. Background noise needs to be well behaved like rain in the trees shshshshshshsh, and it has to be a thousand times quieter than your voice.

There’s a reason the whole second half of the mastering publication is what to do about noise.


Some of the e-learning media that I’ve heard has awful sound quality, but I don’t know how common that is.

My favorite irritation is somebody who reads slowly as for an eight-year-old. Really. If I miss something, I can roll it back.

I suspect it’s a business problem. Nobody is going to retire to Côte d’Azur with textbook readings. So nobody cares.


That was just the answer to the question of there being technical standards. Yes, but nobody seems to pay any attention to them except the audiobook people. And those people are throwing money.

I would still like to hear a sample. Sometimes there are simple things we can do to help.



The Noise_F file is a sample of my noise floor (I left the room to do this).
The Test file is my vocal file.

I use the public Libray’s studio booth to record as it’s free and my basement suite is way too noisey. The only problem is they have the pc set up on the desk so I am assuming most of the noise floor is coming from the computer’s fan.

Other equipment used:

AKG C1000S Microphone
Focusrite Scarlett 6i6 USB Audio Interface

Also, I already ran the ACX plug-in on both of these and I think the readings are inaccurate, it read -61 for the tone of the room and -58 for my actual recording. I had another person check the audio on their daw (not sure which one they used) and they gave a reading of -46. So I’m bit confused about this also!


It’s a lovely voice and I think we may be able to do something about the noise, but please follow the recipe in this post.


Leaving the room doesn’t count. The acoustic signature of the room changes when your body comes and goes—and it’s generally measured when you’re there. The tools work best and most accurately when you have that pause, hold your breath and don’t move for two seconds. Don’t help.

Yes, I think I can hear the computer fan, but there may be ways around that.


With careful noise management, I can get the voice clip to pass audiobook standards.

Diving for noise doesn’t count. If the background noise isn’t audible at normal listening volume, then you pass.

I think you could submit that anywhere and have it succeed.

First three readings and sentence 2/3 down.

Screen Shot 2019-02-27 at 3.48.56.png

ACX Check mimics the analysis tools used in audiobook publication. In general, if you pass that and have your voice, you should be able to publish anywhere. You do have one odd problem. Once you start a job like that, you should stay that way through the whole job. There is no suddenly getting a better studio in the middle of a job. Use the better studio in the next job.

Submit a corrected clip. Processing is more involved than normal because of the computer noise.


Hey Koz,

I have attached the raw recording.



That wasn’t hard.

Screen Shot 2019-03-02 at 15.32.14.png
First three readings and sentence 2/3 down. You could easily read for audiobooks.

If I apply plain Audiobook Mastering…


The performance passes audiobook specifications straightaway, but that’s not what’s in the attached clip…

This clip has an additional noise removal step.

Are you recording in the US? The clip has a touch of noise tone in the background typical of air conditioning, ventilation, or other electrical motor and it has a signature of US wall power (120Hz).

I can improve the clip by either removing the tone by brute force, or apply very gentle noise reduction. With your voice tones, either one will work.

Or, you can ignore it. Your presentation passes with no other corrections, it’s not just as perfect as it could be.


You should not let the mastering publication scare you. The page is exhaustive description of only three tools.

“The captain will notify us if she requires babies and small children scream loudly for the entire trip. Thank you for flying British Airways.”

I was visiting friends and we were getting ready to go out. One of the visitors was a male flight attendant. He came out of the bathroom and did the whole pre-flight in his birthday suit.

“An oxygen mask will fall from the overhead compartment. Place the mask…”

That would be one way to keep passengers from ignoring the pre-flight.


Since you’re not reading for audiobooks, I would probably apply mastering and that’s enough. This is what that sounds like.

If you listen carefully, there’s a gentle hum in the first two seconds (and behind your whole voice). That’s the motor noise. The submission does pass standards like that, just not by a lot. So I think you’re good to go.


Hey Koz,

Thank you for helping out.

What was your additional step for the noise reduction? Was this the step you took to remove the “tone” you were speaking of or is this another step?

And, I record up here in Canada.


I used a gentle, two-step, Effect > Noise Reduction.

Drag-select a half-second to one second of pure noise. This is where that hold your breath for two seconds comes in. Effect > Noise Reduction > Profile. The effect will close itself. That lets Noise Reduction “sniff” the noise to know what to remove.

Select the whole clip by clicking just right of the up arrow.

Effect > Noise Reduction, 6, 6, 6 > OK.

That’s my silly joke Noise Reduction of the Beast. It gently pushes the background noise even quieter without affecting anything else. It’s particularly valuable because Audiobook Quality Control can’t tell you did anything.

That’s it. You should be able to submit to any client with that including ACX AudioBooks.

And yes, you can get through Mastering 25% faster by figuring out where the hum is coming from and get rid of it. No more Noise Reduction.

We also note that you can fail sound submissions for stupid reasons. More than one reader got rejected because they put the wrong number of seconds of room tone before and after their work. You do have to pay attention to the instructions.


One more. Noise Reduction will memorize the noise Profile as long as you don’t close Audacity. The minute you do, it’s gone and you have to drag-select a new profile.


Hey Koz,

Thanks for the tips.

Just two last things that I’m curious about:

Why does the acx tool work better when there is 2 seconds of room tone/noise floor at the beginning? Shouldn’t I just select the part of the track where I am continuously speaking and measure only that since this is what I will be using anyway.

You said when taking a noise profile for the noise reduction tool to take about 1 second. I read somewhere else on this forum that the longer the profile the better.

I know the acoustics of my room change dramatically (because the computer’s fan switches on and off) so that’s why I’m asking :slight_smile:.


ACX Check requires roughly 3/4 second of “clean” background noise or “Room Tone” to give you a good reading. In general, you can’t take that from between words or paragraphs in a performance. If you play the areas between sentences slowly and carefully, they all have you gasping for breath, clearing your throat, shuffling in your seat or clicking your tongue.

During production you’ll wonder how some your production recordings seem quiet to the ear, but they all have stupid-high noise values according to ACX Check. That’s why. There wasn’t enough good quality silence in the performance for ACX Check to accurately measure.

the acoustics of my room change dramatically (because the computer’s fan switches on and off)

Don’t record that way.

Full Stop.

Pretty much no good option.

– Only read when the fans are off.

– Some computers let you force the fans on all the time. We may be able to live with that.

– Separate the microphone and the computer Somehow.

– Build a soundproof barricade for you, the computer, or both.

(Never block the ventilation of a computer.)

– Put the computer out in the hall with long cables (some long cables don’t like that very much).

– Stop using the computer to record.

This is a Zoom H4 recorder. You can put it wherever it’s convenient, it makes no noise, it doesn’t heat up and will produce perfect quality WAV sound files. Substitute the recorder of your choice here. One of the other posters is using a Zoom H2n.

The paper towel roll is from Piggly Wiggly and that’s a heavy furniture moving pad on the table.