I am new to audio editing and am doing my best to learn as I go. My natural speaking voice has very harsh ‘s’ sounds and it is very noticeable in my recording. I’ve been able to get my plosives under control through changing my microphone placement but am struggling to work through how to address the harsh sibilance from my recording. My Into episode is only 10 minutes but future episodes are going to be even longer. I’m looking for either techniques to mitigate during recording or quality plugin tools to remove it from long audio recordings rather than manually trying to edit every little ‘s’. I’m willing to purchase an easy to use and effective deessser plugin if it is going to work but I want to make sure it’s going to be the right one as I don’t have the funds to purchase multiple.
I didn’t have a lot of luck with the existing DeEsser software, but Trebor’s DeSibilator works remarkably well, and it’s a free plugin to Audacity.
Post a 10-second WAV sample of your Essing and I’ll figure out the settings.
It works best after loudness processing and you may be the test case for the Audiobook Mastering Macro. It’s a guaranteed way to get all of your productions to come out the same volume, have them match audiobook volume, and cure your Essing in two swipes.
Koz
You may have Essing problems naturally, but I sorta doubt it. Many home microphones punch “crispness” because it sounds more “professional,” Big Quotes.
It makes my ears bleed and I’m not alone. DeEssing is really popular.
Koz
Here’s my 10 second Audio Clip. I’m using a Blue Yeti mic. Having an easy way to make my episodes uniform would be a dream especially for a newbie like me that also is running a separate business and raising three kids. Making minute edits to every episode is going to be too time consuming for me to do on a weekly basis.
Making minute edits to every episode is going to be too time consuming for me to do on a weekly basis.
I think we can do better than that.
That’s thermonuclear essing. Are you sure you’re using your Yeti right? The early Yeti microphones didn’t tell you how to use them and there are tons of vlogs and podcasts with pictures of people using it wrong (the instructions have since been corrected).
You talk into the side grill just up from the company name.
The natural urge is to talk into the rounded end like a rock band microphone. Resist that urge.
Workin’ on it. I can improve it, but the goal is to make it sound natural.
Koz
You should be in cardioid (heart-shaped pattern) set on the back of the microphone. This helps suppress noises arriving from the rear such as traffic…and room echoes.
Koz
Post back if you’re doing any of these things wrong. That may go a long way to improving the gritty quality.
Did you get the microphone new?
Koz
I think I got it down to your normal crisp delivery.
That better?
There’s two goals. Set overall volume and get rid (as much as possible) of the Essing distortion.
I’m getting the instructions and graphics ready. The first pass through this is going to look like rocket surgery because you have to install and set up the tools. But after that, it’s just Effect > Name > OK, Effect > Name > OK (for example). I’m up to three tools—there’s quite a lot of damage there, but as I said, the settings are sticky and they hang around unless you change them.
I’m doing this in Audacity 3.0.0. If you’re using the earlier 2.4.2, let me know and I’ll check that version.
Koz
This is taking a bit of time because I’m collecting tools and graphics from all over the house where I used them last and I want to get them all in one place when I write this. Simplicity is complicated.
Do post back if you get stuck or I leave out a step. Very strong Coffee/Tea is recommended.
You have to install two tools that aren’t part of the Audacity normal internal collection. Audiobook Mastering Macro and DeSibilator.
– The Macro is a plain text file written in a special way.
ACX-Mastering-Macro.txt (458 Bytes)
Install it in Audacity 3.0.0 by clicking Tools > Macros… > Import > ACX-Mastering-Macro > Open
From that point on, ACX Mastering Macro will appear in that list of available Macros. A Macro is a collection of other tools with the settings already burned in. Some systems call this a “Batch File.” It’s completely automatic.
–Install DeSibilator.
desibilator.ny (56 KB)
This one’s a little tricky. Sometimes a computer download will rename a file like that with .txt jammed on the end of the name. If your machine does that, you have to rename it back with the .txt missing. desibilator.ny is the real full name. This is especially entertaining in Windows which insists on hiding the .txt it added “to help you.”
Right-click the file > Get Info, or whatever that is in Microsoft Speak.
\
Apply The Corrections
Open the same WAV file you sent to us so we’re reading from the same songbook.
Select it with the button on the left.
Tools > Apply Macro > ACX-Mastering-Macro > and it does it by itself. You’ll see the blue waves change size.
Effect > Add/Remove Plugins… > DeSibilator > Enable > OK. This is a one-time-only step to tell Audacity that DeSibilator is there.
Effect > DeSibilator with the following settings > OK. Again, these numbers and settings “stick” and you only have to do this once.
If you can see the blue waves and DeSibilator at the same time you’ll see the little blob at 6 seconds dramatically drop in size without affecting much of the rest of the performance. All your SS words will do that, but that’s the worst one. “That’SS”
But wait! There’s more!
Apply a notch filter to get rid of the worst crunchy distortion.
Effect > Notch Filter at the following settings > OK.
Listen to the to the work and I think it sounds like you, or close enough for jazz.
From this point on…
Select the work.
Tools > Apply Macro > ACX-Mastering-Macro
Effect > Desibilator > OK
Effect > Notch Filter > OK.
That’s it. ACX Mastering will adjust your performance to match Audiobook Standard Loudness. The other two tamp down the Essing. They need each other, so don’t leave anything out.
Post back if something doesn’t appear to work. Those settings will stick unless you change them or something breaks.
Koz > Strong Tea
One note. The notch filter in particular and to a lesser extent DeSibilator are tuned for your microphone and voice. Don’t change anything. There is no getting a different microphone “to make things much better.” The corrections will need to be adjusted if they’re needed at all.
Koz
Rather than use a notch, could use the desibilator on the problem frequency-range, (8kHz-12kHz), with more bands than usual …
May still need a second pass with the desibilator on the (broader) default settings.
Can you get DeSibilator to remember two different settings when it’s used?
I’m fascinated with this particular performance. It has all the symptoms of “ringing” rather than just boosting voice tones here and there. Have you ever had the experience of shouting loudly into a large bell? The bell will hold the tone that it really, really likes whether or not that tone was in your voice. Your voice “hits” the bell. That’s why this performance has one really tall tone at 10106Hz and why the notch filter works so well. It’s not supposed to do that. Something magic is happening.
That’s why formal DeEssing is needed. Normal tone controls won’t do it. This isn’t a plain tonal problem.
I’m building up to a microphone manufacturer’s conspiracy. "Lets make our microphones really crisp and call it “professional.” Also see the “Air” button on a Scarlett Solo preamp.
Koz
Actually, that brings up another DeSibilator question. What are the settings that work well for most people, no matter how long it takes?
Koz
Yes it can have presets …
You’d have to do 2 passes: the high-resolution one first, then the blunter default, (if required).
I heard that too, (particularly after the sibilance was reduced).
If it’s not a object* in the room ringing, then possibly an artefact from echo-reduction applied before the sound got to Audacity.
[* like springs on a scissor-stand ]
You’d have to do 2 passes: the high-resolution one first, then the blunter default, (if required).
Still that’s much better than revising each setting from that scrap of paper you wrote them on.
Universal Settings?
How about lower limit at 3500 and upper limit at 15000? Always. That’s the Fourier peak for human hearing and nobody can hear above 15000.
What’s a good universal number of bands? Even for serious Essing damage such as this post.
That leaves Threshold. I always use DeSibilator after ACX Mastering so the performance volume is constant. You can automate Threshold by sensing the RMS value and using that instead of a fixed, user-selected Threshold value.
That’s the theory that Essing is not loud SS sounds. It’s SS sounds louder than everything else. Sense Everything Else and fix the relationship.
There’s one programming goal, too. If you apply DeSibilator and it’s not needed, it doesn’t do anything. That eases its use in Macros.
Koz
To obtain sufficient resolution to cope with whistles wherever they are (between 3k &15kHz).
IMO would require the 35 band setting, (why I set that as the maximum).
At 35 bands on my Windows computer takes longer than playback time to process.
If it was a 30 second long voice-over, processing taking say a minute is tolerable,
but the brief here was this is an hour-long podcast.
2 passes with the different desibilator settings would be quicker.
A notch filter would be quicker still, but the result would not be as smooth.
Yes, but there’s another hitch in there. The show will not always be processed to ACX standards. So now you have to go fishing for a good threshold at the acceptable number of bands and unless you’re a dab hand at Spectrum Analysis, guessing at the lower and upper frequency limits.
That’s what, four variables? It can’t take longer than six or eight weeks by hand.
That’s why I think automating a lot of this stuff would be a terrific idea. Also instead of going straight for compromised quality, present the decision: I can take an hour, or if you’re in a hurry, I can deliver lower quality faster.
That relives the developer of the decision and automatically delivers the highest possible quality.
A Quality Slider? Ask how many people know what bands are? I can design and build electronics and I’m not sure what bands are.
Koz
The bands are the number of divisions that the frequency-range selected is split into, analogous to bands in an equalizer.
According to Paul L the bands are equal height when viewed on a log-scale spectrogram.
The bands are the number of divisions that the frequency-range selected is split into,
So the More, the Better, the Slower.
DeSibilizer suddenly became insanely valuable to me when I discovered pre-conditioning the sound level through audiobook mastering. That left only two variables, the upper and lower frequencies discoverable through spectrum analysis (Read the Essing Hump) and I ignored the bands.
Boom. No magic settings and far better performance than the other DeEsser tools.
So if you can automate any of those settings it would be a very good thing. Alternately, build audiobook mastering into the DeSibilating process. Now that Mastering is a Macro, that’s not the time-consuming drudge it used to be.
Koz