How to achieve -23 to -18 after normalizing?

I’ve posted the .wav file below good sir. Recording it with blue yeti gain set to the absolute minimum and have made 0 adjustments to the file.

It’s too noisy, but I’m not going to go digging for it. You’re doing the recording wrong.

Forget the headphones for a minute. Set yourself up for a recording and change the Yeti recording volumes and your head spacing until the Audacity recording meters tip occasionally into -10dB to -6dB.

You should be about a power fist away from your pop and blast filter or one Hawaiian Shaka if you’re not using one.

That’s David Greene at the NPR Culver City Studios.

I know the Yeti is a desk mounted microphone and the recommended way to elevate it, assuming you don’t have a mic stand is the book and towel method.

Book and towel has the advantage of isolating the microphone from desk and floor vibration.

Shoot it again.

Good voice, so this should work OK once we sort the technical problems.


Followed the steps for placement and I increased the gain knob to about 45% in order to reach the -10 to -6 db during recording. If it’s still wrong, I think my motherboard usb slots might not be that great for sound recording or the mic itself might be a bunk…My computer does make a slight hum sound, so that might be causing the noise.

I’ve attached the 2nd test file, please let me know if I did it right, I placed my ear next to the mic, and honestly I can hear the computer’s hum which is about 3 feet away from the mic. I’m thinking about returning the yeti and getting a dynamic mic

Aaaaaand. We’re done. I applied the mastering suite just as it appears in the paper.

See first three readings and sentence 2/3 down.
Screen Shot 2017-11-05 at 17.44.42.png
Being obsessive, I did apply very gentle noise reduction of the beast (6, 6, 6) and DeEsser at the default values. I’ll post those shortly.

I expect that to sail straight through both Robot and Human Quality Control—unless you mess up something reading.

Once you establish a reading process, you’re stuck with it until the last page. ACX hates changes in the middle of the book or between paragraphs.

Let me know if you get stuck anywhere.

Even though ACX requires submission in MP3, it’s strongly urged you do everything including performance archive in WAV. Burn the MP3 just before submission. I’ve never read a book but I’d probably chop the performance into chapters or less. ChanningBeerHall-Chapter2a.wav

That’s obsessive again. No spaces and no punctuation other than -dash- and underscore.

I wouldn’t use Audacity Projects anywhere in this process.

Are you reading this for the publisher, or are you the writer?


That’s not to say it can’t be made to sound better. I expect one of the other elves to chime in with better tonal correction. My goal is to get you working.


I think De-Esser is downloadable but I don’t remember from where. I changed the settings to the values posted and the sound is better balanced—less harsh.
Screen Shot 2017-11-05 at 18.18.53.png


Thank you so much, You are a life saver. I am the writer.

It’s not in the effects collection. This is my bookmark.


If this is your first reading, there’s another production effect that may pop up. You get better as you go and about half-way through the book you go back and play chapter one and realize how sucky it is.

It’s not that unusual for New Users to read their first book over again. “I can’t submit this. It’s terrible.”

The production goal is sitting across the kitchen table from someone over steaming cups of hot tea while they tell you a fascinating story. Yes, I know the kitchen is the last place to do a quality recording, but work with me here. That’s what it should sound like. That’s the performance/acting goal.

This is why ACX will kill you instantly if you submit a voice that sounds like a bad cellphone.

Please note we’re not worried about the microphone any more.


And speaking of the Yeti. Can you burn another voice test? We’ll use test2 as a base line reference and burn another, similar test with a dress sock pulled over the top of the Yeti. Make sure it’s pulled down so the grill on top is relatively smooth and your voice only has to go through one layer. I don’t expect much volume change, but the voice may get a little more mellow.

As before. No changes or effects. Shoot it and post it.

Click the pix.



Hey, I do in fact use a foam windscreen cover on the yeti mic. Here is another raw recording, is my voice to harsh recording at that audio volume? I don’t want to embark on recording my entire book only to have it be a waste of time and rejected because my voice was too loud sounding. But if it sounds acx acceptable, then I will proceed. Thank you for your help.

Edit: Hey I apologize, I incorrectly recorded that sample, here it is at appropriate face distance and keeping it at the required db level

Here is another raw recording, is my voice to harsh recording at that audio volume?

It’s not a volume thing. ACX wants your volume to be just so and that’s why they publish the specifications. Within those specifications, the quality of your voice comes down to how much energy is taken up in certain pitches and tones. Generally, you were born with that.

Human hearing peaks around 3000Hz. That’s engineering-speak for fingernails on blackboard and baby screaming on a jet. I wish I’d kept a recording of it, but there was a new presenter on the radio whose natural voice tones would crack glass. They got a lot better over time, but their presentation still fogs my glasses.

But their loudness is exactly correct for the radio.

There’s a cousin to this problem called sibilance. That’s an odd distortion where most of your speech is OK, but all your S sounds turn into ice picks in your ears. That’s the one DeEsser is designed to help. SSiSSter SSuZZy SSinkSS or SSwimSS.

We need to address the foam screen. Was that sold as a pop and blast filter? It will help with that, but it also has the effect of muffling your voice. This tennis racket thing is a pop and blast filter that doesn’t affect your voice.

You should be about one Power Fist away from that for normal presentation. I’m perfectly clear that it’s rough to place the tennis racket when your microphone is sitting on the desk. Probably why they sent you the foam sock. Is there an instruction set with the sock? If your voice does this, do that?

I’m losing track of the versions. I’d like one version with nothing and one version with the sock. Is that test2 and test3?


rejected because my voice was too loud sounding.

You should dig through the ACX publications and see if they let you submit a sample before you read whole books. I think they do now. Everybody got tired of ACX rejecting months of work.

The top four links on this page are from ACX.

You can probably back up a link level or two and find more. They were busy changing them when I wrote the shortcut page.


Here you go, one is with the sock and one without, recorded them according to the distance seen in the pictures as well

While you were doing that, I was pulling test3 apart. There is still a technical problem. There are two places in the dialog where the microphone made up noise by itself.

“Even stars these *** giants of power” 12.85 to 12.90
“Mysterious *** Force” 15.8 to 15.9

In both cases there is a little buzz or “eh” sound that you didn’t make (I’m assuming). Crawl into your headphones and see if you can hear them.

That’s pretty serious because I don’t know of a shortcut to stop it. That kind of error doesn’t fall into a convenient basket of solutions.

I’m guessing it’s a data error.

Is the Yeti plugged into the back of the computer or one of the front convenience sockets? Are you using the USB cable that came with the Yeti?

Are you going direct? No USB hubs, extensions cables or splitters?


I can’t see any difference with and without the sock, so the sock doesn’t appear to make any audible difference. Someone with better hearing can inspect the clips.

That’s the good news. Did you find the two noise errors?


I believe they were user error caused by movement or me not pronouncing clearly, I quickly recorded them earlier so I can find out if my setup is acx capable. Ignoring those two technical errors, am I recording correctly enough to submit to acx?

am I recording correctly enough to submit to acx?

I think so.
Test3 goes through the mastering suite and only misses noise by a tiny amount. Plain Noise Reduction 6, 6, 6 brings it into conformance.
Screen Shot 2017-11-06 at 18.12.57.png
Did you find out the process of submitting a test to ACX?

More importantly, can you duplicate my work on your machine?

We had one poster who repeatedly failed to create a working clip and nobody could figure out why. Turns out they were not using the right Audacity. The custom tools only work on Audacity 2.1.3 or later.


We frown on creating MP3 files unless you absolutely need them, but ACX won’t accept anything else. Do everything in WAV and convert to MP3 right before submission.

I think it’s still true you need to install the “lame” software package to make an MP3 file. Until relatively recently, you could play MP3 files all you wanted, but it required a paid license from Fraunhofer-Gesellschaft to make one. Little known production facts.

Scroll down. You might install the FFMpeg software, too. That’s needed to make and read MP4, AAC and other advanced sound formats.

You should take test3 and see if you can shepherd it all the way through the process. Mastering, noise reduction, editing and creating the MP3 submission file. Attached test3.mp3, corrected, noise reduced and encoded at 192 MP3 quality. It still passes ACX-Check and sounds reasonable, or at least to me. You should critically listen to it.

We had one poster who got bounced by including the wrong number of seconds of silence in the chapters. You do have to pay attention.


Making the MP3 is covered in this video.

It’s one of the links I pointed you to.