Gain vs. Mouthnoises

Hey All!

I’m currently working on my first Audiobook for an ACX client and am having trouble settling on my levels, specifically where to set my gain.

Just FYI I’m using an Audio Technica 2035 with a decent pre-amp and a pop-filter!

I’ve recorded a few chapters already with the gain set to about 80%. I had started out making some samples to try and find my ACX Test ‘recipe’, and since my RMS kept coming in low when I would set the gain to 50%, I jacked it up to 80% and found that plus the following ‘recipe’ would get me through the ACX Test:

RMS Normalizer @ -18
Limiter @ -3.5
Equalization: Low Rolloff for Speech

**one issue to note is that though the 80% plus recipe MOSTLY passes the ACX Check, there are some parts that don’t, due to RMS. These dont seem to be moments when I’m projecting or anything…Confusing.

I am finding, however, that mouth sounds are quite loud due to the higher gain, so I thought I would try recording at a lower gain, about 60%, and run it through my recipe and see what comes out.

I’m finding with the lower gain, two problems:

1-the waveforms are so small in Audacity it’s tougher to edit.
2- once I run it all through my recipe, the result is really loud…and all those mouth sounds I thought were being helped by the lower gain are audible.

Anyone have some insights here?

Have you tried Audacity’s DeClicker plugin ? … Help With Voice Quality Post-Processing (AT2020) - #4 by Trebor

You have the order backwards from the Mastering Suite.

You perform Low Rolloff to get rid of rumble, bus and truck thumping noises before the other tools. Then RMS Normalize to set loudness (I think I used -20). That can produce some troublesome peaks which surrender to Soft Limiter.

https://forum.audacityteam.org/t/audiobook-mastering-version-4/45908/1

I think you’re supposed to measure success by the chapter or segment, not minute by minute. If you take the performance apart in great detail, you may never be able to do anything theatrical or expressive.

The ideal live recording volume has tips of the blue waves roughly half-way or 50%. The sound meter should just be turning a little yellow as it bounces, -6dB to -10dB.

Jury’s out on when to correct for mouth clicks and lip smacks. My fuzzy preference is to do it before mastering because if the clicks are loud enough, it will throw the other tools off. Experiment.

And no, you don’t get rid of lip smacks by turning the microphone volume up and down. They’re always in the sound, you just can’t always hear them. After careful mastering, there they are!


There is a condition where you’re too expressive and theatrical. If there is a wide volume difference between parts of the performance, then you’re into heavier compression and more serious volume controls. You may be out of range of anything Mastering 4 can do. It’s either experimental techniques or try other suites. Mastering 4 isn’t the only one out there, and some of the other ones do feature overall volume setting.

Koz

Seems like a nice microphone, but condenser microphones can be overly crisp and boost any mouth noises you have.

Record and post a little sound sample.

http://www.kozco.com/tech/audacity/TestClip/Record_A_Clip.html

Use some of the words you know you have trouble with and stick to actual narrative, not “testing one two three.”

Koz

Thanks SO much for your thoughts! I’ll get a sample off to you shortly and will run my testing samples through the mastering process in the PROPER order!

Hey Koz- I’m not totally sure what you mean by this? The ACX Check will only allow you to check 10min at a time, so I can’t test a whole chapter at once…

ALSO thank you SO much for your feedback! I will upload a sample soon and I’ll try the proper order in mastering. I’m definitely taking out mouth sounds and loud breaths BEFORE mastering, but I’m running some tests as I go to make sure I’m on track.

I’m thinking at this point that I’ll have to re-record my last chapter as I recorded it with the gain set much lower and though the sample I tried, once through the mastering recipe (in the proper order) made it through ACX Check, I have a feeling I can’t get away with submitting chapters that were recorded at different gains?

I have a feeling I can’t get away with submitting chapters that were recorded at different gains?

The recording levels are NOT that important as long as you don’t clip (distort) by “trying” to go over 0dB. i.e. If you record the exact same thing twice with a 10dB difference you wouldn’t hear a difference after matching the levels in post production.

Your acoustic levels can make a difference. If you speak softly or if you are farther away from the mic the room noise remains the same and your signal-to-noise ratio is worse, And then turn-up the gain digitally, the noise gets turned-up too.

Write down how you did this test.

I applied Audiobook Mastering and the work passed Technical Conformance—even passing noise with good margin.

https://wiki.audacityteam.org/wiki/AudioBook_Mastering


Screen Shot 2019-01-21 at 14.23.29.png
That and you have a good presenting voice. So most of the hard parts are done.

In My Opinion the voice is a little crisp and that’s typical of a condenser microphone, at least the home ones. I’m less good about the theatrical adjustments (and I’m working in the field just now). Other forum elves will be along.

Also My Opinion many home recordings could do with a sock over the microphone. Home microphone makers are convinced that a good crisp sound is the hallmark of a professional recording. It also makes lip smacks thermonuclear.


BetsyAndTracy.XML (454 Bytes)
Here. Try this. Import BetsyAndTracy XML into Effect > Equalization and apply it to that clip. It should take the edge off the sound and make the smacks less obvious without doing anything particularly evil.

Koz

Betsy and Tracy was a show that had the same crisp delivery you have and that’s how I helped it. I saved the setting. It should look a bit like this on the right-hand side.



Screen Shot 2019-01-21 at 14.41.14.png
Koz

IMO the only way to get rid of the mouth-noises is DeClicker software …

Audacity’s free DeClicker plugin is capable of very good results, but it’s slow, and has complex controls.
Paid-for DeClicker plugins are in the range $50 to $150, work quicker, and are simpler to use.

Betsy and Tracy works pretty well. Takes the edge from the hissy sibilants (and mouth clicks) without doing too much damage. This is what other more normal microphones sound like. Gritty, harsh sound is not normal but the microphones are not going to great effort to produce it. It’s all down to frequency and pitch response and you can see it working with the newspaper crunching test.

Betsy and Tracy just removes the boost.

The only down side is two different settings for the equalizer. First pass to get rid of rumble and thunder and the second to get rid of the grit.

Koz

Thanks again Koz! How do I install the plug-in? I’m trying to find that plugin folder I used to install RMS Normalize…but can’t.

Any tips? Such a Luddite here.

This one is handled entirely within Equalizer.

Open or record a short audio. Content doesn’t matter. Equalizer will not work without it.

Effect > Equalization > Save/Manage Curves > Import > point it to the file > Open. The display should be flat line except for that pothole over on the right. I forget whether Equalization then makes you select it again…??.. Equalization is not the easiest tool.

What I expect to happen is a little of the “crisp edge” or “grit” will go away from your voice.

Koz

If you have trouble installing it, post back. I’ll duplicate the steps rather than guessing at it.

Here’s another trick. As an experiment on your submitted clip. Apply Betsy and Tracy twice. See what you think.

Roughly here, we ask you how you’re listening. Good quality headphones or music sound system? If you’re listening on the tiny speakers in your laptop, you could be preparing voice quality nobody will ever hear but you.

I have a joke about being able to hold both of your computer speakers in your hand at once. Don’t bother listening to them. That’s hopeless.

This is also a recipe for confusion. ACX (for example) contacts you to ask about that rumbling thumping sound behind your voice.

You: “What thumping sound?”

Koz

Got it!

I put a sample through twice but didn’t like the effect it had on the tone of my voice, it lost too much of it’s crispness I think. Perhaps just the once it the ticket!

Thanks again! This was very helpful.

Also- I’m using the Sony Professional Dynamic Stereo Headphones, recommended to me by another narrator.

One of the differences between no filter and one pass should be your mouth noises are less obvious. Did you find that worth the effort?

Betsy and Tracy is turning into a reasonable patch between overly harsh and passable without trying to take your voice apart with sibilant patching and other complex tools.

recommended to me by another narrator.

Works for me. I don’t perform so I just pass on positive recommendations from other posters. Thanks.

Koz

The ACX Check time limit is 37 minutes. Not 10. If you run out of poop at 10 minutes (technical term) then you either have a wussy computer (technical term) or you have several Apps running all at the same time taking up memory and resources. I got a 30 minute test track to check just now and I didn’t take a lot of special precautions to get there.

Do you routinely leave everything running on your machine and just bring them forward when needed? One of the managers at work used to do that and he never restarted his machine. The machine would eventually grind to a halt.

Do you have the Application Dock indicators on so you can tell which is running? On my machine, it’s the little lights on the bottom.


Screen Shot 2019-01-22 at 17.01.43.png
I found out recently showing those is optional and I think not having them is a mistake.

Restart your machine and see if the ACX Check limit goes up.

Koz

See?


Screen Shot 2019-01-22 at 18.16.55.png
1860 seconds (next to the bottom) comes out to a 31 minute mono (one blue wave) sound track.

Koz