Audio Sample: Noise Issues. (2.3.2)

I am recording audiobooks on a MacAir (MacOS Mojave 10.14.15) using Audacity 2.3.2 and a Yeti Snowball (brand new out of the box). I’ve previously passed sound tests with this equipment at a different location. I am recording in a new location – high rise apartment – with a lot of building noise I can’t control. I’m in a closet surrounded by heavy quilts on the floor and all sides of me, but not above me. Air conditioner turned off.

I ran

Effect > Equalization > Select Curve: Low rolloff for speech, Length of Filter: about 5000 > OK.

Effect > RMS Normalize: Target RMS Level -20dB > OK.

Effect > Limiter: Soft Limit, 0, 0, -3.5dB, 10, No > OK.

Analyze > ACX-Check

And failed on noise, so I did this: Noise Reduction of the Beast (6, 6, 6) or UNDO and try 9, 6, 6. Most of my files worked with the 6,6,6 and I had two that required 9,6,6.

Then they passed ACX-check for noise. But barely. I was in the -60 to -64 range.

Having previously done five audiobooks that were rejected by Audible (I publish the audiobooks through Findaway), I am once bitten twice shy about my fear of over processing and ending up with a file that is distorted or doesn’t play well on all devices.

Before I invest 80 hours on two books at this location, could someone let me know if in their opinion this file is useable (and how to master it if different from the steps I took above)? If not useable, any other tips that might make this location still viable for recording as I am stuck here for awhile.

I appreciate this so much. Your tips have been a godsend.

Nobody passes noise. Homes and apartments are insanely noisy and most people just don’t notice it—until they have to.

While that’s coming down, can you record a 20 second “clean” sound test?

Go down the blue links. They’re not very long.

This will save a lot of flopping back and forth across time zones.


Before I invest 80 hours on two books

If you’ve been “reading the mail” on the forum for a while you may know you can submit a short test to ACX and save an enormous amount of work.


I had two that required 9,6,6.

There is no treating chapters differently. ACX requires your chapters to match. As an extension of that, they require your chapters beginning and endings match, your chapters match each other and the beginning and end of the book matches.

barely. I was in the -60 to -64 range.

There is a warning about that. You should pass everything at about -64dB to -66dB or better and it has to be very well behaved noise. A noise value of -60dB means someone is going to measure noise slightly differently than you and reject your submission. ACX didn’t publish ACX-Check. That was one of our home-baked tools based on their instructions. They don’t have to measure it the same way we did.

Looking forward to that 20 second sound test.


very well behaved noise.

A recent poster submitted a test that passes ACX-Check, but still failed noise.

What actually happened was they failed Human Quality Control. Their noise wasn’t gentle, spring rain in the trees shshshshshsh, but was instead a high-pitched, screeching, fingernails on blackboard sound. Impossible to ignore. Some USB microphones can do that by accident which is why that clean sound test is so important.


Here you go. Thank you.

Thanks, but we need that two second room tone (background sound) at the beginning. That’s really important for analyzing your performances.

We can’t take effects or filters out of a performance, so we need that raw reading to tell what happened, or what’s happening.

Everybody gets that room tone segment wrong. Hold your breath and don’t move.


That’s exactly what I did. I’ll try again.

I guess I really don’t know what I am doing wrong, b/c it’s actually 3.5 seconds of room tone with me holding my breath, before I speak. I just listened to the wav again. Is there something you could give to me in terms of advice that would identify what you are hearing and why it is wrong? iil just be repeating exactly the same thing otherwise. Is the problem that I need to trim 1.5 seconds off my room and make it exactly 2 seconds, or is it that you don’t hear the silence where I am recording room tone before I speak?


No filters, holding my breath, not moving, just room tone, then I start speaking at 3 seconds. Attached is second try.

Calm down. I think this one is all about me. You posted two sound clips with very nearly the same filename. I got the wrong one. I know that never happens to you, right?

I’m in the field and can’t do detailed analysis, but I don’t have to. Your raw performance is way too quiet.

Let’s see. Snowball. The switch on the back should be in position 1, Cardioid (directional). You should be about Hawaiian Shaka away and speaking into the Blue company name.

I gotta go for a while and play Real Life.


Thank you. I’ll play around with it some more. Have a good day.

I’m back. Remember, we’re all volunteers across multiple time zones.

There are very few ways for a Mac to mess with the microphone volume, but Chat is one of those ways. Do you use Skype, Chat or other communications apps? Do you leave them running in the background when you’re not using them? That’s not a good idea. They change system sound settings on their own and frequently don’t tell you.

There is a user test you can do. Never blow into a microphone, but you can speak and yell as loud as you want.

Can you overload the Snowball? Keep one eye on the Audacity blue waves and bouncing sound meter. Keep speaking louder and louder until the blue waves go all the way up and the bouncing sound meter goes all the way to the right and turns red. You would never do this during a show, but it should work. Overloading the sound channel is valid.

Can you ever get there, even yelling? If you can’t, then there is something wrong and that’s a good place to start. A quiet voice recording invites noise problems.


I had the gain down all the way to reduce noise, but I THINK I’m getting better results now with it turned all the way up and reducing my recording level in audacity to .25 Here’s a sample file of that.

By the way, I am using a Blue Yeti. If I said snowball earlier,I. accidentally lied. I use a snowball for podcasting.

Thanks again SO SO much. My stress level is going down a ton, even though we’re not there yet, I feel better :smiley:

Much better. I can make your last posting meet ACX with just simple audiobook mastering. No noise reduction.

However, you should not be able to change the Yeti volume in Audacity. Did you install any Blue software or drivers?

The voice is clearer, but there is still something going on there. You have a very slight processing honkiness. Are you sure you don’t use Skype or iChat?

Another name for this is talking into a wine glass voice. It’s the signature of noise reduction or environment processing.

Apple (upper left) > System Preferences > Sound > Input. Do you have a setting for [_] Ambient Noise Reduction and is it checked?


Nope, I didn’t install any drivers. I’m talking about adjusting the microphone level on the toolbar in Audacity. Where you drag the little bar left for quieter and right for louder.

OMG, yes, ambient noise reduction was checked!!! It was a system setting I guess. New laptop.

Thank you!

Okay, another test now that the dumb ambient noise filter is turned off! Yay!

also, I don’t have Skype or ichat on this laptop.

thank you!

Cool. The milk jug quality sound went away, but the reduction was doing something useful. I attached a short clip where I intentionally made the noise worse. Is there something in your room which could be making that nose?

Are you sitting your Yeti directly on the desk? The noise could be coming up from the floor through the desk. That’s typical of air conditioners or other air handlers.

Try this. The towel-book method.

Any folded up bath towel and heavy novel. If the noise goes away, it was coming up through the desk and you can record that way forever. If it doesn’t go away, then the noise is probably actually in the room with you. The Macbook Air doesn’t make noises like that, so it’s air conditioner?? Refrigerator??

Now that the voice is clear, I can hear the P-Popping. “Pop Filter” at 8.4 seconds and “Processing” at 14.8 seconds.

Is the Yeti pop filter a black tennis racket like this?

Back away from the pop filter a little more than you are and make up the volume difference by boosting the Audacity record volume a bit.

Home readers never pass noise. All this is perfectly ordinary. The only problem is going to be if you promised somebody a book at the end of the week.


The Yeti has a place for headphones. You can clearly hear P-Popping in your live headphones and adjust your face/microphone spacing until it goes away.

You need heavy, cover-the-ear headphones, not earbuds or thin “fashionable” headphones.

Listening on headphones like that also helps keep your volume more even.

NPR’s David Greene in the middle is using Sony MDR7502 headphones. Those are standard in the movies, too. I don’t like to listen to those for a long time, so I use an older Sennheiser model no longer made.


Luckily, I made no promises on a fast book :slight_smile: I swear I can’t thank you enough. I was floundering. No amount of youtube videos were getting me anywhere.

Here’s one trying the towel/book method, further from the pop filter (yep, I have the tennis racket), and with audacity recording volume turned up a bit. I wore headphones to listen to the previous clip, and I heard the pop. It’s less bad on this new one. I still hear it though :astonished:


Overall much better. Plain, ordinary noise now, no compressors or motors.

I still hear it (popping) though

There’s tricks here, too. The grownups frown on this, but it works very well for me. Don’t speak head-on into the microphone. Push the microphone in an arc to one side or move your chair so the microphone is opposite your cheek, or say, half-way between your nose and your ear. It’s still pointed at you, just not your lips. Most of your plosives (P-Popping) go straight in front of your lips and this adjustment should pretty much eliminate them.

I gotta do a picture of that… It’s the only way I can get through a presentation without constant popping and lip noises.

View > Show Clipping. Set that option and look at the last test you sent. Do you see the red lines in the blue waves?

Screen Shot 2019-10-09 at 15.43.28.png

Those little pieces are too loud and are permanently damaged. So now you’re too loud. The very tips of the blue waves should land at about the 50% (0.5) mark every so often. I think the “5 jillion” sound test was almost perfect.

Screen Shot 2019-10-09 at 15.33.13.png