Audio Feedback request

sharp notch around 8kHz seems to take the edge off.

I haven’t tried a careful notch. I did try approximating the haystack, but upside down. Didn’t work. It made everything muddy.

There was a microphone a while back which would change tonal rendering slightly if you weren’t straight-on to it and you tune vocal tones that way.

threshold dependent on RMS

That’s one problem we don’t have. One of the audiobook mastering tools sets RMS, and DeEssing doesn’t change the waveforms enough to throw off technical conformance, so it can come later.

If you have something that works, then hold onto it. I was startled how well the desibilator works on problem voices. That’s a welcome addition. Trebor also corrected a variable setting in the original DeEsser that could cause Audacity stability problems if you weren’t careful.


“Bob” is a good example of why the default upper-limit of 8kHz on Paul-L’s De-esser should be changed:
most of the the sibilance in “Bob” is over 8kHz and would not be touched without changing from the default setting (of 8kHz).

BTW my suggestion to try RMS as threshold is just that: a suggested starting point.
(If at any point it sounds lispy then the desibilator threshold is set too low).

Thank you all so much for your help with this! I really would be lost without this forum. Is there any option you have to say thank you to kind folk like you? Coffee donation scheme? Beer for Koz fund?

It is sounding a lot smoother and warmer after the desibilator. I think I’m close to a sound I’m happy with. The occasional underlying click which the de-clicker doesn’t remove is a bit of a headache but hopefully that won’t be a problem with future projects where the gain is a little higher to begin with. When I export sample mp3s with the three ACX effects applied, the sound is not as smooth and, though I try to isolate noise between words and use punch paste to cover them with ambient hush, in the outputted MP3 the end of each chunk of speech is more noticeable with a slight click or crackle. Is it worth running noise reduction to get rid of such things or is the only safe way of removing them to go through one by one?

I was also wondering if it’s best to leave the de-clicking and de-essing to the end of the process and whether there’s a recommended order to these things? Should I run the de-clicker and de-esser after I apply the Filter Curve/loudness normalisation/limiter combo or before? If I’m running noise reduction does that go at the end of everything? I plan to leave all that until the end as much as possible to keep the files as pure as possible in case I have further problems to address.

Desibilizer works best after Audiobook Mastering because it uses Mastering’s automatic volume setting.

The other patches and fixes are up to you.

Be sure you’re not Diving For Noise. Evaluation should be setting the show volume for normal entertainment listening and then don’t change anything during the silent portions. Still hear noise? New Users sometimes get to the silent portions and crank the speaker volume up and are horrified at what they find down there. Nobody else is going to hear that.

If you do have clicks and pops, that could be actual computer problems. The machine trying to do two or more things at once.

See: If you spend more than two weeks messing with the computer, stop recording on the computer.


Thanks for all that. I have a bit of a new situation.

We were going along nicely when suddenly, a noticeable constant background buzz started up in the middle of recording. (See the first clip I’m attaching - evil sound.) I swapped out everything else and discovered the problem had to be with the pre-amp (Focusrite Scarlett 2i2) and I couldn’t find a way to get rid of it so I’ve had to change to the only other device I have that will do the job, a Peavey PV8 mixer.

Since then, I don’t find the audio as clean, as shown in the male voice in the second clip. Am I wrong in thinking it’s fuzzy?

It’s our last few days of recording and I’m pretty happy with the female voice for most of the book but don’t want to mess up the last parts we have to do now.

Any help would be gratefully received. Thanks

“evil sound” has gain-pumping: hiss&hum increases when the person stops speaking.
That’s usually due to a compressor working somewhere* which has the noise-floor set too low.

(* compressor could be hardware, (in the audio interface), or software running on the computer).

Thanks Trebor.

So I think “evil” was an appropriate term then. :smiley:

You picked up some sound processing somehow. Chapter 26, At 3.5 seconds, that’s not normal breathing. This passage is:

… natural, gentle background sound, [tick] LOUD BREATH OF AIR, [tick] “David pulled out a seat…”

Something is trying to noise gate and automatically set sound volume to “help you.”

To expand that a little: If your original recording technique was subjected to this kind of processing, that could be where your voice noise is coming from.

I’ll put good money on one of the communications, chat, or conferencing apps is running right now or was running and left it’s microphone processing behind. No question. It’s nice to think you can use a general purpose computer for daily chores and audiobooks, but this is the kind of thing that can happen if you try.

Clean shutdown your Windows machine. Shift+Shutdown (not regular Shutdown or Restart). Wait a bit and Start. Do Not let anything else start. There is no “waiting for a call.” The minute you let one of the chat apps run you run the risk of having its sound processing get stuck.

See if your original recording process starts working.

There is an extreme version of this. Turn off WiFi or disconnect your network line if you’re wired. See what complains. If you don’t know how to turn off your network, this is a darn good time to find out.


Macs have a “Dock” where they collect the icons for your apps and indicate with either marks or lights which ones are running.

Screen Shot 2020-09-13 at 8.28.17 AM.png
Is there a Win10 version of this?



Hi Koz

I’m recording on a macbook with absolutely nothing else running on it. Wifi was connected but there’s nothing else open in the dock and it’s in a user account we set up especially to record. I opened Activity monitor to check on background processes and there’s very little running. No sign of Skype or anything like that.

I’m editing on my windows 10 PC. It couldn’t be an app messing with the sound after it’s been recorded presumably. I’m most concerned about the male voices but I’ve experimented with levels on the mixer and that seems to have helped with the roughness of the sound. I’m attaching one more file with the current setup without any editing done to it. Do you still hear the problem? Is the voice too harsh? Admittedly it’s an angry part of the text from an angry character and please ignore my bad acting, it was just a level test.


I didn’t complain about that earlier but yes, the man’s voice is much louder than the woman’s which is distracting. You have to get stress and anger with acting, not volume. You can’t do it with post-production editing and special effects, or at least not easily.

freak.wav doesn’t seem to have any damage. No switching or obvious volume shifting. The voice seems normal. I hear normal shirt and pants shuffling and breathing sounds, and if I master it, it easily passes ACX-Check.

It couldn’t be an app messing with the sound after it’s been recorded presumably.

I wouldn’t assume that for an instant. This effect is coming from somewhere and if the Mac is clean (and it seems to be), then there’s only one candidate left. I would be a lot more soft and fuzzy about this if the damage wasn’t textbook chat or other voice processing.

I would be looking for a different machine to cut and master the chapters. You can come back and fix this machine later.

And if I wasn’t Debby Downer enough, have you ever heard of an audiobook with two different voices? I haven’t.

You should worry about stepping over the line between audiobook and radio theater. ACX’s metaphor is someone telling you a fascinating story over cups of tea. Not two people telling you a story over tea. If for no other good reason, it does some entertaining things to rights, ownership, and permissions. Some of the same reasons you can’t use music.


Yes, the male voice is far too high compared to the woman’s. The male part isn’t recorded yet so I’ll make sure that it’s set lower, as I suggested before.

Dual narration is a thing actually and some people seek it out. It’s probably more common for romance novels but I thought it would work well for mystery too as the characters are quite big and bold. I have seen a number of them on Audible, including mysteries. I believe it’s called duet narration when a man does all the male voices and a woman the female voices. It wasn’t our original plan but, for a number of reasons, it seemed like the best option.

I’ll change computer for the editing then. The reason I was using the PC was that I could hardly hear the sound coming out of the macbook. It’s very quiet but also seems to smooth over a lot of inconsistencies.

Thanks for all the help again. 3 Days left to record the final chapters with the main narrator. Fingers crossed it goes to plan.

I believe it’s called duet narration

Writing that down.

I could hardly hear the sound coming out of the macbook.

I can’t either. My Air is much better at theatrical sound than the MBPs. The Air doesn’t use tiny speakers. It uses compression drivers under the keyboard. I think it’s a cousin to how smartphones do it.

Normal people would be cutting on headphones (not earbuds).


Sorry, I meant how loud the sound was coming from my macbook to my on-ear headphones, not the tiny built-in speakers (I’m not that much of a newbie!). I ended up buying a half decent audio technica pair but, connected to the macbook, it was still far too quiet and presumably there was some kind of compression on it - as you say, it’s similar to what iPhones do as I sent the chapter 26 clip to my brother who couldn’t hear anything wrong with it on his Apple devices but it’s clear as day through my headphones connected to my PC.

I plan to shut everything down on the PC, open a new user profile and ensure that there are no interfering background apps to avoid anything interfering with the sound whilst editing.


Since you’re doing a Windows start-over, Windows has patches, tools, and effects, too. Speaker symbol in the lower right. Dig down and turn off anything trying to “help you.” Playback Enhancements, Noise Reduction, etc.

Google your brains out.

Screen Shot 2020-09-14 at 7.17.00 AM.png

connected to the macbook, it was still far too quiet and presumably there was some kind of compression on it

Mac doesn’t do renegade compression…yet. But it does do multiple different places to set volume.

Apple (upper left) > System Preferences > Sound > Playback. There’s a volume slider there.

Web pages can have their own volume controls.

Apps can have their own controls.

Desktop > Upper Right. There’s another volume control up there.

Screen Shot 2020-09-14 at 7.27.34 AM.png
And there’s no shortage of controls inside Audacity.

Speaker symbol up in the menu bar.

Screen Shot 2020-09-14 at 7.31.35 AM.png
And each timeline has one.

Screen Shot 2020-09-14 at 7.31.51 AM.png
That last one is dangerous because it can make you think you have a far different sound mix than you actually do. Leave that one alone.

My office machine doesn’t use any of the Mac services. I turn everything all the way up and use a separate hardware HiFi volume control into external amplifiers and speakers.

Screen Shot 2020-09-14 at 7.41.08 AM.png
Software controls, as a fuzzy rule, are volume decreasers. If you turn everything up (except that timeline one), you get a volume control ratio of 1.

I’ve never had trouble getting headphone volume on any Mac, which makes me think you have something set wrong.


Sorry, missed the obvious one. Laptop keyboard F10, F11, and F12. Poit Poit Poit Poit.


Ahhhhhhhhhhhhhhhhhhhhhhh! I was so sure I’d done all that stuff within Windows but, on double checking it now, I discovered a Dell sound app running on my PC that I have never opened or have any recollection of installing and it had this Maxxaudio option selected - It wasn’t affecting the recording but it explains why I found that everything had a slight blown out quality even at low levels. I think it was probably taking the mono file and playing it back in stereo to supposedly enhance the sound.

Cheers for that. I am an idiot.
Screenshot 2020-09-14 162234.jpg

I have a story about getting a company PC from the Systems People who checked it out as OK. No matter what I did, it would not pass “clean” sound. Turned out it was running “Cathedral Effects” in the playback chain.

If this was easy, anybody could do it.