AudioBook Mastering version 4

I’ve been keeping the laptop just outside the closet door on a stand

That’s one of the home recording conflicts. Keep the noisy computer out of your recording room and watch the screen and don’t put the computer more than one USB cable away from the microphone. It’s not an accident that quiet Macs show up in successful studios.

I wasn’t recording with the headphones on

If you have a recording system that allows real-time monitoring with headphones, that can make a big difference in self-correcting voice volume.

Classic telephones (remember telephones?) had “sidetone” where a little of your own voice was fed back to your ear as an aid in self-correcting. Cellphones can’t do that because of digital delays. There was about a year or so where people got used to talking into a dead piece of plastic.


In the picture, I think that’s a pre-amp on the left (black device). Any recommendations there? What’s the orange thing to the right?

Correct. That’s a Behringer UM2 preamp. It adapts a full-on XLR analog microphone to USB. It supplies 48 volt phantom power (if needed) and may be used for monitoring as well as “Perfect Overdubbing” with a built-in headphone mixer. It natively mounts mono which is perfect for Voice-Over and Overdubbing. No post-production conversions.

The orange thing is an external USB hard drive. You should be able to point to two different devices or locations containing your valuable work. If your internal hard drive clicked a couple of times, choked and caught fire right now, where is your work?

I have a significant difference of opinion with the ACX demo people. They insist it’s a terrific idea to record directly to the external drive. I don’t agree. My history in video services proved repeatedly and painfully that’s an awful idea. Record to your internal drives(s) and then make copies to the external.

The Behringer isn’t the only one in this class. There is a Focus Scarlett Solo which does pretty much the same thing.

There is an odd problem with these devices. The upper-end stereo devices (Scarlett 2i2) won’t gracefully handle a single microphone. They are not mixers and they insist on recording a stereo show with your single microphone on the Left (or Right depending on where you plugged it in).

You can sometimes force a mono recording, but you’re subject to overload, undervolume, noise or mix problems if you do.

I generally use the next step up from even this for commercial recording. Same analog-type microphone, but plugged into a small field sound mixer and then on to either a USB converter or a computer with a top quality Stereo In. This is a broadcast sound shoot I did.


Hi Koz & friends,

Newbie here, who has been looking around for answers. I’ve found most of what I needed, but am still unfortunately failing the noise floor requirements from ACX Check. Not sure what I can do at this point. I use an Audio Technica AT2020 USB (I know, not ideal) and a pop filter, and I am in a very quiet room, I think. I’ve read that using Noise Reduction isn’t recommended, but when I DO try it, based on the specifications you list in this thread (the 666 and the 966), then THIS clip passes, but other sections of my recording fail. Any help is much appreciated.

Screen Shot 2017-12-01 at 6.04.00 PM.png

Post a 20 second sound test on the forum according to this formula. Try to hit it around 18 to 20 seconds. Some posters recorded just “Hello” and stopped. That’s not long enough for us to tell what’s going on.

Do Not Change Anything. Record it, Export a WAV and send it. If you’re counting, the file size should be under 2MB.

Everybody fails noise. Most people live in crazy-loud environments and they just don’t notice it until they record and have to listen to it played back. “Is that what my apartment really sounds like?”

Some of the noises are just not your fault and many of them can be helped, not always with Noise Reduction.

You need to be on Audacity 2.1.3 or higher. That’s not optional. Some of the tools fail on earlier Audacity versions.

I don’t think there’s anything wrong with the microphone, but I would have bought the analog version and then plug it into a Behringer UM2, UMC22 or Focusrite Scareltt Solo for the USB connection.


THIS clip passes, but other sections of my recording fail.

One of Human Quality Control’s requirements is that voice quality of all your chapters and segments match. Once you apply a correction anywhere (Noise Reduction), you need to apply it to each chapter forever.

This is why people go to some effort to make sure the voice is recorded pure and clean at the beginning. Yes, it’s possible to force some microphones to work when they don’t really want to, but then you have to force each chapter for the whole book or publication. You can do that if you want. I’ll record with a good microphone in a quiet room.

We’ll need to see what the problem is.


Hi Koz and all, Here’s a raw, unedited sample following the instructions you posted. This time, I added more padding to the room (a walk-in closet). The computer was a few feet away from the USB mic and I also turned the lights off. Typically, I would read the manuscript or text off of an iPad, but I didn’t bring the iPad in for the sample. Sorry for the short clip … this was under 2 MB.

Just thinking out loud, but for those needing to keep a noisy computer out of the recording area, has anyone tried using a projector to display the computer screen on a wall?

has anyone tried using a projector

Have you found a quiet projector? They all have the universal requirement of keeping the light source cool, much easier with LED sources, but still.

Extending the keyboard/mouse and display to an outside computer is not a bad way to go.

I’m finding that my paleolithic technique of printing my script in large type on actual paper is not shared by others. Either a Pad or Phone should both work as neither makes a lot of noise in use. I do note that you can’t be a clumsy buffoon and do that. One tap in the wrong place, your script goes away and you get to edit your performance. Also there’s the problem with phones that they do radiate radio energy and can get into your microphone. They also ring.

There are some interesting ways to mark an edit. If you fluff, compose yourself and continue with the performance, trying to do work in even sentences. Do Not leave corrections until hours later and try to shoe-horn new words into an old performance. The question becomes how to mark the error so you can easily find it later. Tap a wine glass? Leave a big gap?

I need to leave for a while.


I use an iPad to read from, and run the mic to the laptop which is on a stand just outside the walk-in closet door that I record from. I can open the door, press ‘record’, close the door, and record without the laptop in with me. I’m thinking about running an external monitor into the closet to I can keep an eye on recording levels.

As far as recording a mistake, I just snap my fingers and then repeat the word/phrase correctly and continue on. That makes it easy to find the mistake and delete it.

I just snap my fingers and then repeat the word/phrase correctly and continue on. That makes it easy to find the mistake and delete it.

Writing that down…

Do you have any lamp dimmers in with you or Compact Fluorescent Lamps? Regular tube-type fluorescent bulbs? There is a significant portion of the noise at 60Hz and 120Hz typical of a US location. Lamp Dimmers can make noise with tungsten (old fashioned) bulbs. Many of my CFLs create a buzzing sound. I’m phasing them out. LED lights are usually desirable, or run tungsten bulbs at full brightness.

Candles work.

Are you on a laptop? can you run it on battery power and disconnect the shore power adapter for a test?

Many consumer USB microphones take shortcuts with noise prevention and allow sub-sonic noises to get through. Yours does that. Yours has significant noise tones down at thunder and earthquake rumble—but of course, you can’t hear them. The Noise Measurement System can hear them and they can mess up a recording “for no apparent reason.” Unless you know where to look.

The first step in the correction suite, Equalize > Low Rolloff for Speech can get rid of most of it, but there’s still stuff left over.

I’m on a larger machine now and I’m ripping apart the noise to see if there’s anything else that can be done. I couldn’t make noise conformance, either.


I got noise conformance with a comfortable margin, but I had to work hard to get it there. Attached.
Screen Shot 2017-12-17 at 10.10.49.png
Next time you submit something, you can convert it to mono. It uploads faster, takes up less room and we don’t care. ACX requests submission in mono. Drop-down arrow on the left > Split Stereo To Mono. [X] delete the bottom track.


I went in with a hatchet and deleted three different tones. 60Hz, 120Hz, and 138Hz. 60 and 120 are “simple.” Those come from power in the US. 138 is not simple or normal. That almost has to be coming from a fan or motor somewhere.

From inside your booth, can you tell if your computer is running? If you left instructions for someone to turn the computer off while you were in the booth not watching, could you tell when it went down?

I applied the notch filters before anything else.

Effect > Notch Filter: 60Hz, Q10 > Enter.
Effect > Notch Filter: 120Hz, Q10 > Enter.
Effect > Notch Filter: 138Hz, Q10 > Enter.

Then the rest of Mastering Suite 4 and then Noise Reduction 6, 6, 6.


You can use your microphone as a diagnostic instrument. The AT2020 responds best from the front, the company name side. Aim it at your desk or other lamp and see if the 60 and 120 goes up. Move it closer to the door with the computer behind it and see if the 138 goes up. If you’re listening on headphones, you can actually hear many of these effects as you move the microphone similar to that guy with the metal detector at the beach. I don’t remember if the 2020 has a place for headphones.


If you do the diagnostic thing, “slate” what you’re doing as you go. “This is my desk lamp.” _______________. “This is my ceiling lamp.” _______________. “This is the closet door.” _________________.


Are you using the little three leg stand that comes with the 2020USB?

Tripod desk stand with folding legs for secure and easily portable tabletop use

It doesn’t say anything about shock mount or desk/floor vibration suppression. There’s an easy way to provide that.

The blanket over the desk is a pretty good idea, too.


In larger systems, vibration suppression happens up near the microphone. That’s the spider of rubber bands.

Screen Shot 2017-12-17 at 13.39.02.png

You don’t have to do it up there, but it gets more expensive and awkward if you do it anywhere else. This is for, say, air conditioning or water pump noises coming up through the floor.


And you can tell ahead of time if you have noises in the desk or floor with a mechanics stethoscope. In English, that’s a screwdriver. You press the handle against your ear and the blade against the table or floor. Mechanics have been using this trick for ages to find bad engine parts—with the engine running. Yes, the first four pages of google hits turn up actual stethoscopes, but on the fifth page, I found a guy using the screwdriver.
Screen Shot 2017-12-17 at 13.46.33.png
Isn’t noise management just the most fun?


The info at says to apply noise reduction multiple time for different types of noise. So would it be feasible to apply it once to get rid of a low level of hiss, then perhaps again to the track to get rid of breaths or some other type of noise that occurs a few times in the track (lip smack, etc)? If I do it once with the beast (6,6,6) by using a section of the track with just the room noise/low hiss, will it also reduce those other things, or make them more obvious?

Sounds lovely doesn’t it? The answer to most of that is no.

Noise Reduction doesn’t work reliably on moving and changing noises. So there goes gasping, mouth ticks, buses going by and planes overhead. TV on in the background is guaranteed to kill your show.

Noise Reduction will try to reduce whatever you put in the Profile step. Drag-select some noise. Effect > Noise Reduction > Profile. Now select the whole track and Effect > Noise Reduction > choose the reduction settings > OK. The profile will stick as long as you don’t close Audacity, so you can use it again and again.

There are purpose built filters and tools to address the most common home recording noises. There’s one for Hum, one for Yeti Curse, one for Sibilance, etc. The trick is to figure out what you have.

Plain noise reduction of the beast (6, 6, 6) is an invisible reduction that can make an OK production into a good production with no other apparent damage. More importantly, ACX can’t find it. They hate Noise Reduction.

You can get stiffer. 9, 6, 6 works but by the time you get to 12, 6, 6, you can hear voices starting to sound wine-glassy and cell-phony and you have to stop.

We should remember in all of these instances the tools are Step Two. Step One is not record the noise.

It’s grand to think I can set up a Yeti in my kitchen, record audiobooks, make a fortune and retire to a cottage on Côte d’Azur. But probably not.

People still convert cloak closets into studios to avoid most of the hassle of post production. Everybody fails noise. We live in, and largely ignore a very noise environment.

So recording in a dead room is Very Highly Recommended. It’s not unusual to need to record only after about 10PM or so. I have to do that because even though I have a terrific room, I also live next to a busy street and the room has a window.

One April First I posted an effects filter which would turn the garbage you shot into a studio production. It was a big hit. I need to see if I can find it.

There is one interactive problem I know of. I can’t prove this but I think stiff noise reduction makesss sssibilanccce worssse. So if you already have a very bright and sparkly reading, Noise Reduction may push you over into harsh, gritty and screeching.

And we warn readers constantly to Export WAV (Microsoft) 16-bit sound files of any original readings. Move them to a safe place before you mess with copies. Cloud or thumb drives work.


Here it is.

I’m pleased to announce the first public “beta” release of the Professional Audio Filter (PAF).

Some of the controls and installers are a little rough yet, and on occasion it’s been known to crash in a shower of brightly colored confetti, but I think it’s worth getting it out there for wider testing.

The object of the filter is to convert whomever you shot in the field and in whatever bad conditions to a quiet, polished, professional-sounding voice track.

Intensity and phase adjustments are provided to suppress “white” or microphone self-noise, proximity effect, competing voices, street noise, engine roar, large and small auditorium echoes, air conditioning compressors, vent whistle, and within reason, clipping, crosstalk, cracking, and overload distortion.

Whimsey is alive and well in Los Angeles. We are preparing an announcer module with pitch, sibilance, and depth controls that we’re calling, for obvious reasons, ‘LaFontaine.’ “In a World…”

I’ve been working with the programmers here in LA and friends at other audio web sites and forums to pull together a lot of different efforts. I aim to be “Welch.”

If you remember in LZW Compression, Lempel and Ziv were the world-class data compression Subject Matter Experts with no organizational or people skills. Welch was the one who got them both into a room and said nobody was leaving until they generated an integrated product. And he had a gun. Nothing like adding Smith and Wesson to the team to polish those people skills.

We’ve included installers for all the major audio software packages: SoundTrack Pro, Audition, Audacity, ProTools, AVS, etc. We even got one to work in the older, revered Cool Edit.

It’s open source on all three platforms.


Koz: I’d appreciate a bit more clarification of your recommended mastering process: EQ, RMS normalization, Limiting, and (if needed) Noise Reduction effects in turn. FWIW I use Audacity v2.2.2 under Win10.

– Room tone: ACX requires that “Each file must have 0.5 to 1 second of room tone at its beginning and 1 to 5 seconds of room tone at its end”. Must these room tone segments be “raw” (without applying the listed mastering effects) or can/should they reflect application of these effects?

– Noise reduction (NR) effect :
o (a) If RMS normalization raises the track’s volume relative to the raw recording, do I understand right that it will raise the noise floor as well?
o (b) Does a higher noise floor used as the NR noise profile yield greater resulting noise reduction (all else equal), or is extent of noise reduction controlled entirely by the NR effect’s Step 2 Noise Reduction (db) setting?
o (c) Bottom line: Does it make any difference whether the NR effect noise profile uses the pre-normalization or post-normalization noise floor? If Yes, how/why so and which noise floor should be used?

Thanks for clarification including pointing out any misconceptions in these questions.