Meeting Acx requirements

Hello…Koz…are you there?? I’m sure by the time you read this I will be in Acx prison. I was so excited a few weeks ago when I submitted 4 Auditions and got 4 offers. I’ve stalked your every word and tried so many combinations and I have failed. I received my first rejection and more to follow no doubt.
Enough about me… Ha. Can we be friends? Smile.
Audacity 2.0.6 exe
VISTA 64 bit
Lexicon Alpha
Shure SM58 Dynamic Mic
—I have a Harbinger Mixer but Audacity (or my computer)crashes when I try using it so it’s unplugged…
Mic & speakers plugged in to lexicon which is plugged into computer & my "computer speaker has to be plugged in to the lexicon headset spot or nothing works. Ugh.
Audio Host/MME
Playback/USB Speakers
My Acx rejection stated the Noise Floor was the problem and the files were heavily processed…
I’m thinking of getting a Tascam & Condenser Mic but it won’t help me in time before they come and cart me off… My Analyze-Contrast will not work so I’m in the dark to even read the audio. When I read the Amp after recording it shows 50.
Thank you in advance for reading my long list of blah.

My Analyze-Contrast will not work

Do you know what a “doorknob moment” is? That’s the doctor’s office when the patient, after having received a glowing good bill of health from the doctor’s inspection, turns with their hand on the doorknob to leave and says, “I bleed out the noise every night for fifteen minutes. That’s not important, is it?”

It might be.

Some of the advanced tools are only available in Audacity 2.1.0. You should upgrade.
New Link Below

Come back and yell at us if those links fail.

Analyze > Contrast only works on mono tracks, or it did last time I used it. If you’re submitting stereo tracks for a vocal performance instead of mono, can you come up with a crystal-clear, super convincing, iron-clad reason?

You can create a test by splitting a stereo performance to mono using the drop-down menu under the black arrow to the left of the track.

Split Stereo to Mono. > [X] Delete one of the tracks.

Select the whole track by clicking just above MUTE. Does Analyze > Contrast work now?

After Audacity is running and stable and we’re all on the same page, we can address the sound issues.


You should have Steve’s rumble filter if you don’t already. It may or may not be useful, but it’s good to have it in the event it’s needed. It’s used with Equalization’s Length slider (to the right) all the way up.

Attached: LF_rolloff_for_speech. Unzip and install it in Effect > Equalization.

Instructions should be buried in that link.

Koz (326 Bytes)

I appear to have the link wrong.

It’s being changed as we go.


Thank you in advance for reading my long list of blah.

The long list is welcome. I have a good feel for what you’re doing and a good head start on figuring out how to fix it. It was also a good thing that you included the details of what ACX found wrong with it.

Once you make a noisy recording, it’s very difficult to pull yourself out of the hole. The tools for noise suppression and the ones for performance volume management tend to fight.

“My RMS is perfect, but now my noise is too high.”

It pays stunning dividends to make a good quiet performance at step one.

There is a sister posting from someone trying to make a good, quiet sound file out of his performance. I forced it to work using seven Audacity tools and filters. When I got done, it would pass ACX, but it still didn’t sound very good and could have used more work.

Post Production is not supposed to be a career move. I recorded a simple performance here and with two simple filters and volume change, brought it into ACX compliance. If I changed the room I recorded in, I could leave out one of the filters.

One of the other posters can do that, too. Stop talking and press Stop. Touch up here and there, cut it to the right length and out the door.

Grathwahl said in one of the ACX videos (highly recommended) If you need to apply tons of corrections, you’re doing it wrong and even if you do manage to create a desirable product, it takes you a week to do it and you’ve missed the marketing step.

But we’ll see what happens to you when you get your system together and post a test.


Hello Koz!! Thanks for the information. I only record in Mono. I did download the new Audacity version but could not. Get it to work. Should my settings for audio host, etc… remain the same as stated in my first pleas for help? Ha. I also uninstalled Audacity 2.0.6 before I downloaded the new one… Right ?? I will try again & address your instructions.
Ahh yes…hand on Doorknob… Just got my 2nd rejection- same problem “doesn’t meet Noise floor requirements & overly processed with noise reduction making it difficult to understand what is being said” It didn’t seem to matter that it sounded awesome on my media player before submitting it though… Surprised?? Nope.

We should get Audacity 2.1.0 running. The tools, instruments and filters are superior to old versions.

ACX’s reference for clear, natural sound is an unprocessed, level-corrected submission. Yes, it is possible to make that and by using non-magic, normal equipment. But possibly not most USB microphones. Jury is still out on that one. We used to claim USB microphones were the “get out of jail” devices, but enough people have had troubles we don’t say that any more.

Once ‘unprocessed’ is your bar, most sound processing sounds funny to you. The clipped, slightly hissy, speaking over the inky blackness of space with no breathing vocal sound. This kind of thing is called out multiple times in the ACX training videos. Take whatever steps you need to record with a good, clean microphone in a quiet room. Noise Reduction/Removal is not welcome.

You can get good recordings in some strange places. One poster was having award-winning room, slap, echo, and noise problems and then suddenly she didn’t. She dragged her equipment to work which turned out to be a very large, partially padded, acoustic ceiling office space over a weekend. It was stunning. Suddenly the sun came out.

It’s only a matter of time before I try a simple recording in my overstuffed garage. If you ignore the labels on the boxes (Donna Summer 12" 1976) It turns out to be a remarkably quiet room limited only by the metal sliding door. As I am also the doyenne of furniture moving blankets, that doesn’t bother me.

I’m including Steve’s Rumble Filter. If you get a new Audacity version, you may need to install it again. It works with the Equalizer “length” slider all the way up.

So let us know how you get on with the 2.1.0 install and which Windows you’re using. I’m not a Windows elf, so if you need helping setting your sound devices I can make sympathetic noises and do very little else.

Koz (326 Bytes)

I’m reading the Lexicon user manual. You should have the first knob line1/Inst all the way down and the second knob, Line 2/Mic all the way up. It should be possible to lean forward into the microphone, yell loudly and get the red peak light to come on as a test. Or get somebody else to yell. Or clap your hands hard right in front of the mic.

If somebody gets the light, then the microphone and Lexicon are working normally. If nobody can get the light to come on, then we need to find out why. An SM58 is completely capable of creating an overload condition as a normal thing and without damaging itself. It’s a rock band microphone. That’s what it does.

If the light comes on, then run Audacity and make sure your volume control next to the little microphone symbol is all the way up (attached). This is the place where our two versions are going to be different if you’re not in 2.1.0 yet.

Also make sure your Audacity is connected to the Lexicon and not some other service. Again this may be different from yours.

This is a listing of what I did to your earlier clip to force it to pass. Audacity 2.1.0 is required.

Peter Piper Processing
Audacity 2.1.0 on the supplied clip.

– Effect > Normalize: [X]Normalize to -3.2 [X]Remove DC > OK.

– Steve’s Rumble Filter.
Install and apply LF_rolloff_for_speech.xml
Effect > Equalization: LF_rolloff_for_speech, Length = 8191 > OK

– Noise Removal.
Drag Select 1 sec to 4 sec. Effect > Noise Reduction > Profile
Effect > Noise Reduction > 12, 6, 6. > OK

– Effect > Compressor: -20, -50, 3:1, 0.2, 1.0. > OK

– Effect > Normalize: [X]Normalize to -3.2 [X]Remove DC. > OK

RMS (loudness) is a little odd in this clip. Drag-select just the spoken parts (9 through 20) and measure RMS. That passes. If you select the whole clip, the vast quantities of silence throw off the measurements. That’s why we recommend two seconds of room tone on these tests and not the much longer one that ACX wants.

Screen Shot 2015-04-29 at 18.15.27.png

Hello again Koz!
So, I fell into the Abyss…crashed everything then reset all with new Audacity version, then my microphone refused to work …but I have two identical Mic’s so i attached the “new” same type Mic and sheesh… All is working. Note: I’ve changed my Gain settings, mic placement…since the attached KOZTEST1 (catchy name right) I am also attaching KOZTEST2 of my recording environment now. I knew that unless I get as close to pre-mastering “Ideal” numbers as possible…I will never succeed…So, I did that and I STILL cant apply the right effects to get the magic numbers… You know …right Peak , Wrong RMS…or the flip of that…ugh. I PRACTICALLY HAVE TO SCREAM TO GET THE RIGHT PEAK HIT…like that. smile.

In the first recording the microphone gain is way too low, but then I suspect you knew that.

The second recording should meet the ACX standards with only the amplitude corrected… “Effect->Amplify”, set the new peak level to -3 dB.

You could probably turn the gain up a bit further than where you were running for the second test. If Audacity shows an adjustable slider for microphone level I would try just setting that to the maximum (1) then use just the knob on the front of the Lexicon unit to get the right level. If you watch the Audacity meters you should be peaking at about -6.

I’m pleased to listen to your works, but not right now. I have to go out and play Real Life.
A different elf may drop in. My world tends to be the Live - Capture - Microphone end of it, so you may wait.

I’m working with flynwill to Figure This Out. It should not be that hard to buy a microphone and make a recording.

Yes, I know. There should also be world peace, etc.


Oh my g… shd i get excited yet…Happy CInco de …The Last recording i sent to you was with Gain “Full on” to the right (On Lexicon) Monitor Mix Knob (on Lexicon) Full Left & i turn it to the right for playback after recorded. The far right knob “Output Level” I’ve tried in every spot but not sure if it plays a part in my problems yet…

My chain on completed files has been (since new set up) per an acx persons advice-Limiter -7, Norm-3, High Pass 200HZ, I tried that 9million times but awful results so I did High pass first then Norm then Limiter bc I read that you should only apply a Limiter as the last effect so I tried that too…no luck.

However, I tried your prescription (below) :frowning: Minus the Noise reduction) out of fear ha— & all of my numbers were in ACX Compliance!! yeah.!

Effect > Normalize: [X]Normalize to -3.2 [X]Remove DC > OK.

– Steve’s Rumble Filter.
Install and apply LF_rolloff_for_speech.xml
Effect > Equalization: LF_rolloff_for_speech, Length = 8191 > OK

– Noise Removal.
Drag Select 1 sec to 4 sec. Effect > Noise Reduction > Profile
Effect > Noise Reduction > 12, 6, 6. > OK

– Effect > Compressor: -20, -50, 3:1, 0.2, 1.0. > OK

– Effect > Normalize: [X]Normalize to -3.2 [X]Remove DC. > OK

I thought that I had to apply at least 3 effects to get a Retail quality, Polished audio,wrong page again…Thank you Koz!!

I thought that I had to apply at least 3 effects to get a Retail quality

No. I did my sound test with no complex effects. Steve’s filter, a little hum rejection, set the volume and out the door. That’s what’s supposed to happen.

Flynwill noted outside the forum that he thought your latest effort would achieve ACX compliance. I like “retail quality.” Can I use that?

As a chance observation, I noted that women’s voices tended to come up short on the RMS loudness test. It’s been notable, but maybe not remarkable. It makes electronic sense. If you have a wispy, shy, woman’s voice, you’re not going to put as much pure horsepower (to mix metaphors) into the sound track as, for example, I do. My bass male voice can go through soundproof walls. You have a higher compression ratio than I use for most corrections. It’s normally 2:1 for many people. Also, women tend to need compression at all more than men. This makes a noisy microphone a very serious problem. Your presentation quality and the microphone noise are close to directly competing with each other.

I’m going to listen now, but flynwill says even though the whine sound is still there, it’s way down in volume.


Yes. Very minor corrections.

“Effect->Amplify”, set the new peak level to -3 dB.

I don’t do it exactly like that. I use Normalize.

Effect > Normalize: [X] Remove DC, [X] Normalize to 3.2 dB > OK.

This will seem odd, but 3.2dB correction will take care of you accidentally exceeding -3.0dB when you create the MP3. MP3 can change sound levels and ACX automated testing will flag anything that goes over -3.0dB. Too low but very close should sail straight through.

I found Normalize handy in multiple places in the correction process and those exact settings are appropriate or harmless no matter where it’s used. We like simple.

I like the presentation voice.


Another production note. Your RMS loudness in that clip is bumping right along on the minimum needed even with the compressor and corrections. You will find that your RMS setting will go up with fewer and shorter sensuous pauses in the work. This is messy now, because I really like it the way it is.

To give you an idea, there was a recent test posting where the performer left huge silent gaps before and after the actual performance. Overall, it was OK, but the only way I could pass RMS was to drag-select and measure only the vocal parts and ignore the silences. Silences drag down the overall average (sensuous or not).


Thank u Koz…I try to focus on sounding professional because my natural breathy sound can cross over into a different section of the “Bookstore” haha I need those pauses though…its part of the story telling…but i will cut them down somewhat to help my cause.
When I do the 3 measurements and the numbers are good…that isn’t all there is to do, correct? Isn’t there always a “mastering” chain or effects to apply to achieve “Retail Quality” (yes you have permission to use it)- or is that only when the numbers are off? forgive me, it hasn’t really sunk in that i could see a light coming through…

About my Microphone, Shure SM58, should I get a different Mic? I will still have the same voice…but it would be nice if I could find that sweet spot on the Mic that can hear me easily. Weird note, I opened the door on my closet…ummm I mean recording studio…it was hot in there…and continued recording as a test - I didn’t have to bob my head around or be sooooo close to the pop screen to find the loud spot on the Mic

it hasn’t really sunk in that i could see a light coming through…

The light is coming through. Neither flynwill nor I agree with the ACX video where the presenter applies so may corrections to a performance that the blue waves are unrecognizable. I don’t think that’s valuable in a submission.

That corrected clip… shucks, I didn’t post it !@#$. should pass ACX automated testing and your natural presentation voice should take you the rest of the way through.

In my opinion. But we’re starting to accumulate a list of people that are now getting paid for their voice work, so it is possible to make this happen.

On a side note, Ian in another thread just in the last couple of hours created a voice track that will, with very minimal corrections pass ACX compliance. I know, right? It amazed everybody and scared the cat. No, you don’t need multiple passes through noise reduction or careful applications of Eye of Newt® if you shoot it right.

It’s not rocket surgery, but you do need to be clear you are playing the part of the recording engineer. It’s not automatic—buy a nice mic and rake in the checks.

I personally would stick with what you have for a while. If you start juggling hardware, you could put yourself back at the beginning of pass after pass until you got all the corrections stable again. And you may never get them stable. In another post, they are struggling with a SM7 instead of an SM58. It’s a different microphone and it’s not as loud. You know the problems you’re having getting loud?

In the production videos (not the engineering ones) they stress continuity of performance.

Whatever you do, it should be possible to do it through a whole book or publication, including announce in the broom clos…, sorry, “Studio.” Nothing takes the listener out of the story faster than a sudden change in sound quality. I can get Ian to tell you how he does it in his broom clos…, sorry, “Studio.”

I think you’re up to the step where you read about the actual mechanics of submitting work if you haven’t already. Do you have a file manager company lined up? Drop Box?? Email will only typically hold 25MB at a time, so that’s not going to post all the Wuthering Heights chapters. You’ll be in MP3, so that takes a little of the sting out of it, but still. Also, you’re uplink speed will suddenly become important. That thing where you said, “I only care about watching movies and videos on line, so the uplink speed isn’t a big deal.”


Your words:

“No. I did my sound test with no complex effects. Steve’s filter, a little hum rejection, set the volume and out the door. That’s what’s supposed to happen.”

Me: A Sound test on the KOZTEST2 clip that I sent to you? I know Steve’s filter, normalize to set the volume-
High pass filter for "a little hum rejection " ?

& No. I don’t have a file management company. It’s already been a problem. I tried Google Drive & it didn’t work but stated the server I was sending it to rejected it-- there’s that word again.

Also, do u have your Coffee or Chai Tea …almost finished-
Regarding Continuity, I record as many chapters on one recording as possible. I save an unedited WAV , import that Audio to Edit & Export a “Selected Audio” of each Edited chapter to “Master” separately. Pheww. Am I on-track…ha I like each chapter in its own neat little space. Does that cause a problem with how overall continuity behaves.
And Thank you flynwill for your help!!

So if I’m following this correctly you are currently recording with all the stops pulled out and knobs set to 11. That is you’ve got all the gain knobs set to their maximum. If that’s the case, so be it, you can add the additional 7 dB or so in “post”.

It also sounds like you are working very close to the microphone (as in less than 4" or so), as you may have already noticed when you get that close it gets a lot harder to stay a consistent distance. Every time you double the distance between your mouth and the mic the level will drop about 6 dB. So if you’re working at 4" and then lean in 2" accidentally you’ll get 6 dB louder. If you can back off to 8-12" then that 2" wobble will have a lot less effect. Koztest2 has a noise floor at about -77 after I adjust the level to the -3 peak ACX wants. So you currently have a 17 dB margin on your noise floor relative the ACX requirements. You can probably afford to loose 6 dB of that to back off from the mic a bit if it gives a better performance.

Also, hopefully you already realize this, but I feel obligated to point out that the SM-58 is a “end address” mic, meaning it should be pointed at you. (There are mics that are side-address where you talk at the side rather than the end).

I was going to comment on the dramatic pauses as well. In the little clip they are just fine, but when you sit down to read us “War and Peace” (or perhaps “Fifty Shades of Grey”) those pauses should become a lot fewer and farther in between. Unfortunately we don’t really know how ACX goes about those measurements, and to what extent their software measures or ignores the silent bits. We also don’t know for sure how they go about measuring your noise floor and whether or not that requirement is “A Weighted”, so we assume not. (A-Weighing is common when measuring noise, it means that the signal is first filtered in a manner that is thought to mimic typical human hearing response. A-Weighted numbers can be anywhere from 3-8 dB better depending on the spectra of the noise.) Koz tells me that he did ask for more information along those lines but was turned away.

I agree that you should stick with the SM-58 for a while, you are getting perfectly fine results. At some time in the future you may want to invest in a large-diaphragm condenser mic, but a good one of those is going to set you back $300-400. Because they have an amplifier built into them they are usually anywhere from 12-25 dB more sensitive. (The internal amplifier is necessary because the signal from a condenser mic is extremely weak compared to a dynamic mic like the SM-58). However, they also need something called “phantom power” to provide power for that amplifier, and it appears that is a feature your Lexicon amp does not have, so you’ll be looking to replace that as well.

In your first post you mentioned you owned a “Harbinger Mixer” but that it caused Audacity to crash. Can you tell us the Model?

All that and I did notice an inconsistency. According to your spells and incantations, the last thing you did was Normalize. And yet. The clip you supplied was a bit off and I “cured it” by applying Normalize (and flynwill by Amplify, a sister tool).

Sound clips don’t change volume by flying through the internet, so something isn’t strictly right.

And no, we don’t know how ACX does it, but the last set of interactions with them seems to indicate they measure noise with no conditioning or changes—no A Weighing. That simplifies things because our Audacity tools will give roughly the same number that they get. However, that also means a very common sound problem, sub-audible thunder/earthquake rumble is enough to throw the measurements off. That’s where Steve’s filter (LF_rolloff) is valuable.

My home-recorded clip has hum in it from the Hum Monster that Lives In My Attic. So I had to apply a notch filter to get rid of it. I have neighborhood high tension power lines over the house and almost without question that’s where it’s coming from. Most people won’t have that.