Meeting ACX Requirements w/ Audacity


I ended up making the meters bigger, they’re definitely a bigger help as to viewing how my levels are, THANK YOU.


As for sound, I set my input volume to .70, it looked like it was set at that on your picture so I figured I try to give it a go at that setting, and this is what I got. PLEASE disregard the popping, it will be fixed soon.

I take it the “anything over 0dB” sound was that screeching noise at around 12 seconds in the clip. OMG that was horrendous, but I felt the need to keep it in there for all the constructive criticism.

Seriously, thank you for your time!

I felt the need to keep it in there for all the constructive criticism.

The worst thing you can do is post a clip that you’ve already tried to patch and failed.

You make my postings sound better than when I write them. How do you do that?

I think we’re better than 95% home.

I bet you can’t hear those pops. This is where trying to mix, patch and correct on tiny speakers at home isn’t useful. You may hear screeching and odd gritty distortion at 12 seconds, I hear a bass drum, explosive, fourth of July detonation that’s enough to tip over the wine glass and terrify the cat. I expect a call from the neighbors. I bet it registers at the CalTech Seismology Lab.

2.1 Richter.

That’s the obvious thing to fix. That should fall to the pop and blast filter (black tennis racket). Even that might not be enough. You may have to speak into the microphone slightly to one side or with the microphone slightly high and pointing downward so the air associated with the P sounds doesn’t go straight into the microphone.

I like your presence and clarity. I can’t detect any room ambience or echo at all. My normal joke is that I can tell the size of the room you’re standing in just by analyzing the echoes and reverb; I can’t do that with you.

One thing you didn’t do was leave enough dead-air “room tone” for us to tell what’s actually back there. Start the recorder and say something innocuous so we can get a voice level and then hold your breath and stop moving for a few seconds. The performer is the room.


Agreed, I heard the same glass shattering noise you did playing it back in Audacity, it was atrocious with headphones! :open_mouth:

Thank you for your kindness, I just like to story-tell… ACTUALLY I’m used to “one-takes” due to my time long ago in a radio station, doing live traffic reports and on-air commercials, so I got that going for me, but I have to much life in me to be announcing the morning traffic reports to the general public, so I’m trying to branch out somewhere else. :wink:

Alas, pop filter will be in my grasp today, and I will see about tinkering with the mic in various positions. The new recording I just posted I have the mic facing downward and I’m speaking over it. Just trying it out.

I’ve also left you all with the “room tone.”

I know you say you don’t notice room echo and such, but when I playback, I can hear my voice echo in the dead room. Am I going crazy?

OK, you’re recording a bit too hot now.

Where you wrote “I’ve put the input volume on .30 and got these results. Am I getting there?”
That was just a little on the low side.

Your most recent sample is clipped in places, and it does not appear to be just “wind blast” imho.

The Goldilocks level is probably a bit closer to the first than the latter.

Noise level is not bad, but has the curse of USB - electrical interference from the USB power. Fortunately that can be substantially reduced with either the Noise Removal effect and/or careful filtering. We’ll get round to that later. Achieving the best possible “raw” (unprocessed) sound is the first priority. “Post process” should be about adding a bit of polish to a good recording rather than salvaging trash (not saying this is trash, just emphasizing the priorities :wink:)

Yes, I can hear your voice echo.
How far did you say you were from the mic?
The closer you are to the mic, the more direct sound the mic will pick up, so room echo will be relatively less.
Are you using a pop filter? (

You make my postings sound better than when I write them. How do you do that?

The answer is people pay you to do that. Did you have a board-op when you were working? That explains your unfamiliarity with the sound meters

You should not be looking down on the microphone. The Blast Zone is straight in front of you and down. So if you’re going to launch your Ps, that’s where they’re going to go. Slightly up or slightly to one side should be OK.

In this shot…

You have to know that when the performer was seated, his nose was almost level with the primary microphone (the gold thing). The other microphone is at even more of an angle. Both walked away with a good track.

This positioning, in addition to working out acoustically, happens to leave your desk free for copy and in this case, making phone calls to coordinate the shoot.

I haven’t heard the track yet.


I can hear a little echo in there now. Also I don’t think that’s our friend the USB frying mosquitoes. I think that’s computer fan noise, but it’s so low, I think gentle noise removal should do it. I applied correction at about 13 seconds. Attached.

Now resolve the popping and I think we got it licked.


Did you have a board-op when you were working?

Why yes, yes I did have someone working the board.

Here’s a new sample of it in my sound booth, and me facing directly at it, with my nose almost parallel to the mic.


That’s impressive. I’m having to work to find anything wrong with it. Is that speaking through your black tennis racket, or did you just move the microphone up?

Please note most of the peaks hover around 50% with occasional peaks a bit over and nobody hits 100% anywhere.

No odd sounds, crashing S, sibilances, or piercing overtones.

Very slightly high noise levels, but they’re the mixed variety with no one tone predominating — and fixable.

I need to go searching for the ACX Compliance forum thread. I don’t know we ever formalized it into a wiki.


You are the wise-ass, comic-relief voice of the spaceship computer in the new interstellar action procedural on HBO.


That last posting very nearly passes ACX peak volume specifications just sitting there. 2.8dB. The spec is 3.0. Easily correctable.

One serious restriction of Noise Removal is that the Profile step where you tell Audacity what noise you’re trying to manage has to be in the same environment as the show. You can’t, for example, sample that Room Tone clip you provided earlier and accurately apply that to the last clip. If you moved the microphone or changed the room (and you did, right?) you need new Room Tone sample.

More as I figure it out.


The clip I tested passes the ACX RMS volume spec — barely.


I got Steve’s Noise Floor Tester to reveal -50 noise floor. In an earlier clip, I successfully applied a 12dB noise reduction, so that puts the delivered noise floor at -62.

Let’s see. What else? Those were the biggies, right?


In English, I believe that with a little gentle noise reduction, that last clip would pass ACX Audiobook Compliance. Not by much and there will be an insane urge to apply tools and corrections just because they’re cool, not because they’re required.

And it’s our perception of Audiobook specifications. They could still shoot you out of the sky for a minor infraction, but I think you have enough voice presence to roll right over that.

And they all but said so. If Stephen King decided to read, I don’t think anybody would reject his work for a noise floor problem.


Are you up for a multi-hour read? Novels last longer than describing traffic on Atlanta’s Connector.


Boy did I love reading the updates you gave :slight_smile:

That was with just moving it up and no black tennis racket!

I didn’t actually “move rooms”, I just moved my mic into the mobile booth I DIY’d up.

If the ACX req. say to “hover” around -60dB, is -62dB going over? I’m sorry I’m not hip with the lingo yet.

I’m glad you believe we (because you helped me) can do this!

I’m totally up for long reads, my voice isn’t the problem, it’s getting the tinkering with the equipment right, and by George I believe we’re almost home, I shall be reading your eloquent comments with the pop filter shortly.

You’re a savior! :smiley:

Painting with verbs.

One of the baseline understandings for digital audio is that the numbering system seems to go backwards. Zero dB is the reference point. If you try to go louder than that the digitizing system runs out of numbers and starts eating the show. Blood everywhere. There are exotic exceptions, but that’s a good starting point. This damage is called clipping and you can’t recover.

Working down in volume, your vocal peaks should be around -6. The average overall loudness (hard to measure) is -18 to -22. That’s the overall drone of your voice as compared to crisp spoken consonants.

Noise or Room Tone is expected to be -60. I have never shot anything that quiet in my life and I don’t expect anybody else has either. I think your last clip had room tone noise in the -45 to -55 range (which is slightly louder). Remember I claimed I could hear your computer running way in the background? I ran a gentle noise reduction to artificially push it quieter to the magic -60 point.

Nobody hits quieter than -60 by accident. That’s a full-on studio or a room where you can’t sleep because it’s too quiet. The farm in Upstate New York where the nearest neighbor was 1/4 mile away was like that at night.


How did you achieve your studio?


I guess the next action item is to record a clip from your new studio including a 3-second Room Tone gap. I’ll take it through the ACX process and see where we get to.

Try not to gulp or shuffle (or check your FaceBook) during those gaps. Last one I could hear you doing something in the background. Audacity will try to remove anything in that sample (Profile) from the show. It’s important that it be pure room.

It’s harder than it looks. During that conference room shoot, I had to force four people to be absolutely still for ten seconds (per the client request).


The horse’s mouth is here:

I’ve not had time to listen to the most recent sample, but previous samples had a fair bit of low frequency noise that will push the rms measurement higher than is really justified by the sound (we are using unweighted rms measurement). Rolling off the sub-bass will give a more realistic measurement. (sorry, I’m in a bit of a rush but I’m sure koz can fill in details about sub-bass rolloff.)

Thanks for the help Steve!


Here’s the clip, with the black racket attached! :3

Let me know what you think!

Once again, I’m sitting straight in front of the mic, and I left you some “room noise” at the end of the clip. I haven’t changed rooms.

I’m excited to hear the critique!


I achieved my studio by being crafty and just working for the money.

koz can fill in details about sub-bass rolloff

Sub-bass is the rumble of a large truck going by your house that you can barely hear but ripples the wine in the glass and scares the dog. Also see: 2.0 Richter earthquake. Steve designed a pre-baked equalizer filter that can be applied to get rid of much of that rumble without harming the voice.

that [rumble] will push the rms measurement higher than is really justified by the sound

Which will require volume boost compression which will then screw up the noise floor (room tone) measurement.
See how darn much fun this is.

It could be argued I think, that after the original pass, we should stop looking and let ACX catch it. If they consistently catch it, then we’re measuring it wrong.


So once we figure out my end and record my first bit, I should send it in and see what they say?

It might also be argued that since you’re talking through your tennis racket with your microphone elevated, you may no longer have a rumble problem. Popping P sounds generate significant low frequency rumble.

In the event you’re picking up rumble through your floor or desk, that’s why shock mounts or spiders are made.

If you’re handy with plumbing supplies and rubber bands you can make one. Here’s one design for a Shure SM-58 microphone.

It’s just jammed together. No glue.

Saturday’s are a little busy for everybody.