Passing RMS on Audacity 2.1.1

Woo! I’m so happy. Thank you. I ran an ACX check with no processing and it passes noise, but not RMS. The mic has volume and sensitivity controls (I have everything set to mid-point). It’s also set to a cardioid pick up pattern (also has bi or Omni directional). Is the cardioid best?

The company I’m doing work for now wants their clips really short - like .1 seconds of room tone before the audio starts. Is there an easier way to set that in Audacity rather than manually clipping each file?

If New Mic is the G-Track, then that may be all you need. I got very low noise level and a passing ACX Check noise of -69dB. No noise reduction.

Screen Shot 2019-12-23 at 18.15.30.png
This is after just the three mastering tools. It took me 20 seconds. Everything passes. No other corrections. Given your recent history, we need to see if you can keep doing that through a whole book or a job.

If you’re just steaming for voice correction, try DeEsser at these settings after mastering.

Screen Shot 2019-12-23 at 18.25.19.png
You’re listening for sharp SS sounds to go away.



Cardioid (heart-shaped) points the sensitive part of the microphone to you. Omni tries to record the whole room—all directions, and bidirectional records you and whoever is behind the mic but not the sides. Speak into the company name.

Is there an easier way to set that in Audacity rather than manually clipping each file?

Nobody can hit that kind of timing without editing. You’re also going to have to suppress the breath you take at the beginning.

“[Gasp] Hudson Valley farms…”

So cut out the gasp and then cut off everything to the left of 0.1 seconds.

I’m guessing you’ll be the voice of a machine or process.

“The eight forty-five from Leominster is now arriving on track 14. Please mind the gap between the train and the platform.”


Yes, it’s work for an education company reading Question stems and answer choices (just like my new samples). I installed the de-esser you posted earlier, but my settings screen looks different from yours…
Screen Shot 2019-12-23 at 9.46.08 PM.png

And yes, it is the Samson G-Track.

I suggested Steve’s De-esser plugin,
Koz was showing Paul-L’s DeEsser plugin for Audacity.

Paul-L’s is capable of surgical-precision, but can be slow to process audio,
whereas Steve’s can be much faster, but less precise.

BTW if you are using Steve’s these are the settings I’d recommend with the new mic …

suggested settings for Steve's DeEsser with new mic.png
Apply these de-esser settings after your audio is loud enough to pass the Audacity ACX test.
( Steve’s & Paul-L de-essers are threshold-dependent: if the RMS is low they will have little effect).

We’re not curing killer problems, here. There are just one or two places where the SS sound is slightly more pronounced than it probably should be.

Just before 3 seconds

And just after 5-1/2 seconds.
Cacti’SS Needles

If you didn’t notice it, chances are good nobody else will, either.

Who’s writing the book? I don’t think I would have used the word “cacti,” that nobody else on earth uses. That and I don’t think the possessive is warranted there. Cacti Needles are a thing and they don’t separate. It’s like describing Cacti Green. “Cactuses” is acceptable plural in my dictionary, but that gives you plu-possessive punctuation nightmares. Is it Cactuses’ or Cactuses’s?

“While washing my car, I accidentally backed into my cactus’ needles.” (Actual, real-life event).


Thank you for the settings.

In this recording gig, I’m recording answer choices – so some of them are wrong…

With the very short (.1 seconds of room tone) clips that they want, I’m finding I’m passing noise easily (-84 to -87 dB) when there is 1 or 2 seconds of room tone. When I run the check again though, I’m failing (-24dB)…??? It’s driving me crazy. This wasn’t happening before (Or maybe I wasn’t re-checking after I trimmed?)

Here is a raw sample of one it’s happening to.

Sometimes the Audacity ACX tool gives spurious results, depending on the audio selection.
If you view the waveform in dB, (rather than in linear),
the room-tone RMS level is below -60dB after raising the overall RMS to -20dB …

room tone level on dB scale.gif

So, does that mean I am actually passing and it’s just giving me bogus results? Is there a way to fix it? I did probably ten clips last night and not one problem. Tonight, I haven’t had a single one pass…

Thanks for your help, Trebor.

Yep, the ACX tool is giving me bogus high values for noise if I select part of your audio, rather than all of it …

ACX tool can give bogus results.gif
Your audio does contain a lot of infrasound, which is inaudible, but can be measured by RMS tools.
Even if that’s not causing the spurious results, you should remove it with a rumble-filter, or bass-roll-off filter …

high-pass filters which remove infrasound.gif

Ok. Where should that go in my sequence? Now, I’m doing,

Filter Curve (LF Rolloff for speech)
RMS Normalize

Should I do the rumble filter in place of something, or in addition? And where in the sequence?

Also, does it matter if I master before or after I trim to .1 seconds at each end?

My studio is in the basement- could that raise the infrasound?

Sorry - I think I see. Do the 100hz rumble, then LF rolloff and etc…

I think it’s noise generated by the electronics, rather than vibrations in the studio.

Either 100Hz rumble, or LF rolloff will suffice: you don’t have to do both.

On the subject of Low Frequencies, IMO you’d benefit from boosting the bass, (100Hz-300Hz range) …

Audacity has a bass-treble effect which you can adjust in real-time: i.e. when the audio is playing.

Either 100Hz rumble, or LF rolloff will suffice: you don’t have to do both.

Exactly. Low Rolloff is part of a carefully selected suite of tools. A harmonious grouping. I know it seems too simple, but it’s impressive all the things those three tools are doing for you. There are postings from people determined to get it down to one tool or step and it usually doesn’t go well.

Some USB microphones create errors down in thunder, earthquake and rumble sounds. Most times it doesn’t make any difference to the show so nobody worries about it (and it’s expensive to fix the microphone). The mastering tools do care, so Low Rolloff gets rid of rumble errors and should come first.

You have a job that doesn’t fit into the audiobook format. ACX Check needs 3/4 of a second of clean room tone to measure noise. From your description of the job, that’s the last thing they want. You can do all the mastering and measuring first, get a top quality show, and then cut it down to their format. Once you cut it down, measuring your quality is going to be much harder. I think you have to do that step blind.

If you do either of the De-Essing steps, they should come after mastering. DeEssers need stable quality and perfect volume to work right.

I’m not an editing genius. I can think of insanely complicated ways to cut the piece down to 1/10 of a second before the first word. I think I’m going to stand back and see how others would do it. Is there a specification for the end, too? Is it the same 1/10 second?

It’s further complicated by not being able to use the first ‘real life’ 1/10 of a second. That’s where your lip smacks happen. There’s a tiny one in your NewMic.wav posting, so you have to get rid of the smack first, then produce the 1/10 second. See that tiny-teensy blue thing (red dot) just left of your larger blue waves, “Select the possessive…” .

That’s a lip smack and you should cut it off before you do any other cutting. It’s almost impossible to avoid a gasp or lip smack just before your first word, so whatever we decide, I think you’re going to be doing a lot of it. I can’t believe there isn’t a shortcut or a fudge somewhere. Doing it the formal editing way is a career move…

As we go.


Please note with your revised Rag Studio, new microphone, and application of mastering we’re not worried about making RMS or suppressing audible noises any more.


No. Wait. I got a better one: “Ragamuffin Studio.”

I processed the phrase “Select the possessive noun below” from your posted sound file.

Drag-select just the word blue waves.

You can use the magnifier tools to zoom into the words so you don’t have to hit it inside a sea of other words.

I only ever use just three tools. Zoom into a selection with Command+E. Zoom out a little bit with Command+3 and zoom out to the whole show with Command+F.

Note the selection does not include that tiny blue tick to the left. You can play just this selection with Spacebar Play to see if you hit it without cutting off any sounds.

Isolate just those words from anything else on the timeline.

Edit > Remove Special > Trim
Tracks > Align Tracks > Start to Zero

There is no shortcut. I asked.

Home Key > Generate > Silence: .1 seconds > OK.

Select (on the left) > End key > Generate Silence: .1 seconds > OK.

That will give you the words with a tenth second of silence before and after.

File > Export (Whatever their delivery format is.)*


Yes, that is a cheat. The 1/10 second is not Room Tone but Audacity generated Silence. I don’t think they will care. It’s not there long enough to hear the difference.

  • I can see storm clouds. You can’t submit in MP3. MP3 is not a precise editing format and will screw up the 1/10 second of silence. You can’t do production in MP3 either for sound quality reasons.

WAV (Microsoft) 16-bit works. When you get a final with all the silences in the right order, Export a WAV as your production archive no matter what the client wants.