Something else that confounds me ...

First of all, take a breath. I was in the same boat as you. This is going to be real simple. I just made a video today on how to “Meet the ACX Requirement, EVERY SINGLE TIME”. That is the title of the video. If you do as I say, I will fix your problem, (99.9% Sure), within 10 min after you reply back to this thread, if I am still on-line. All I need to know is this.

  1. What are your input levels for your dynamic mic, according to Audacity?
  2. How far are you away from your mic, as you record?
  3. What is your max noise floor, when you start to record?

If you can not provide me with this information, there is no way I can help you. Watch the video, answer my questions and believe it or not, you should be able to solve your own problem. If not, please post your answers to my questions and we will go from there. You can start the video at 8 min 52 sec and stop at 12 min 30 sec, to get the answers I need from you. At this point the video is only 4 min 30 sec long.

@Dan_Tucker. Not all my books. Just the two I mentioned, and a couple of text books on English, and one on currency trading. :slight_smile:

@DLVoice. Sorry to hear about your problems. I have heard back from the ACX sound engineer, and my 25 minute sample I sent is is good to go, as soon as I remove the mouth noises. Just a bit of editing.

Seriously, I think you may be over correcting. I have a -70 thereabouts noise floor to start with, Mrs in the -20s, and peaks below 0. All I ever need is normalise + limit, or just Normalise. Rarely maybe Amplify, but really only once I think. I now have 4 chapters done. 13 to go.

I’ll maybe post the Sound Engineers comments later.

Wish I knew a way to get rid of mouth noises in production :slight_smile: but from Googling, it’s a common problem.

A lot of mouth noise is due to “Dry Mouth”. Here is a great article from the Mayo Clinic that deals with this subject. I am breaking about 10 of their rules as we speak! :stuck_out_tongue:

Not Dry Mouth in this case. Copy of a post in Stackexchange. Interesting post actually. Taken from this discussion. narration - How do you lessen mouth noise in VO recordings? - Sound Design Stack Exchange

Okay. Pardon me if I sound ranty.

I promised myself I would not ask this question, but frankly, I am fed up with it.

I record narration as a main part of my job in sound.

I have recorded possibly over 3,000 hours (final product, edited down) of narration, ADR, overdubs, etc. etc. etc. in my career.

I have still not found how to lessen someone’s mouth-noise in the recording.

I have searched and searched and searched for a remedy to this.

I have read Randy Thom’s article about it.

Many people’s answer is MIC POSITION. This is utter rubbish of advice. I know well that the moment you add that top back on a voice that you lose by going off-axis that those clicks are just clear as day in the recording and have to be edited out, so I know for a FACT that doesn’t work. Not to mention the recording sounds horrible in the end because you have a U87 pointed at your ear. Sure, you can even move the mic 4 feet back from the talent, you’re going to have one thin recording and a lot of room to battle then - especially if it’s supposed to be narration.

Lemon water. This has had mediocre results for me…

I’ve tried grapefruit juice,

I’ve tried having the guy suck on lemons,

I’ve had the guy put vasoline on his teeth,

I’ve had the guy eat so many green apples he was hypnotized into thinking he was Johnny Appleseed,

I have forbidden the consumption of all coffee,

I have forbidden the consumption of sugar,

I have forbidden the use of honey and other saliva-producing foods,

I have told the guy to drink water - funny, everyone asks the talent to do this when he gets mouthy and IT JUST MAKES HIM MORE MOUTHY, Surprise!! You’re just putting more wetness in his mouth!!!

I have also tried Izotope RX and I personally think it adds digital artifacts to the recordings and makes the voices sound dull and processed…

I have tried everything I could possibly find on the internet or from other professionals about this and they all have had no avail.

I personally think it’s an awareness thing. I think that the talent just has to know what it is and learn for himself how to fix it.

But, what have you used in the past that has actually worked?

Is there a “magic pill” that someone can take and MAGICALLY he has NO mouth noise and won’t need ANY editing at all?

I highly doubt it, but I’m working on 20 seconds of narration right now and getting it clean as a whistle and I’ve spent the last hour on it.

One hour for 20 seconds of voice…

My standards are pretty high for this sort of thing as you can probably tell…

But besides that, what have you found that has worked for you.

Has it ever been a problem with your production and has a project ever been rejected back to you saying “It’s got too much mouth noise in it”?

Sorry for ranting but I just don’t think 20 seconds an hour is very viable.

Thanks - Ryan

The only way I’ve found to fix it so far is by the hunt&delete method.

Now that I’ve had a night to calm down, I started playing around with it again. What I did this time was to go through the longest track and delete a lot of the dead space in between my talking.

Let me back up and explain that days ago, I went through several of my first tracks and inserted room silence noise in between my lines (deleting the extraneous noises in between (such as when I had to stop speaking because my family was walking around upstairs, or when I had to swallow, or clear my throat, or whatever). Some of these pauses were long, and I used Audacity’s punch copy/paste feature to insert normal room silence in between. I wasn’t deleting these long pauses, just inserting silence. I’m thinking that those long stretches of silence were somehow responsible for what happened when I would run various effects. So today I deleted a lot of those long, empty spaces, leaving a more uniform looking track (like how it would typically look when it’s properly edited). Then I ran low-pass filter, noise reduction at 6,6,6, compression (this time at 2:1), normalize, and nothing vanished from my track (from what I can tell so FAR); I also passed ACX.

I’m itching to go through and make my edits NOW because that’s going to be the longest, most tedious part of this whole process, and I’m running out of time. But I think I’m still not fully understanding something. I believe it was Steve who said that if I needed to remove DC offset or reduce low frequency noise, I needed to do that first or it could cause clicks at edit points. What if I don’t need to remove DC offset? How would I know if it’s necessary? I’ve been doing it automatically with the Normalize feature, but what if it’s not needed in my case? Or what if low frequency noise isn’t an issue for me either? Would removing these things when it’s not necessary cause issues? BTW, my 8K whistle problem is no longer visible in my tracks, so I haven’t even been running the 8KNotch fix anymore.

Also, what is involved with “post editing processing and mastering”? Would this be things like click removal or de-esser or limiter?

I’ve written down what I think the rough process should look like for my situation (I have to keep making lists to keep my head straight about this), and here’s what I’ve been referring back to lately (please let me know if I’m WAY off base on this):

  1. Remove DC offset/run low-pass filter
  2. Noise reduction at 6,6,6 because I was told (I think by Steve) that NR is supposed to be done before using the compressor (note - this may not even be necessary; my room silence noise isn’t that bad … perhaps I’m just being picky)
  3. Compress 2:1
  4. Amplify/normalize (I prefer to use normalize) to -3.2 (with DC offset UNchecked)
  5. Run ACX check
  6. Edits (cut/delete, fix clicks, plosives, and loud esses)
  7. Post editing processing and mastering (does this mean just running the limiter in my case? Or is there something else I’m missing?)

ROFL :smiley:

Another slightly less drastic, but no less arduous technique, is “hunt and filter”.
One of our developers is an audiobook narrator / producer, with a fastidious approach to mouth noises. He became so fed up with the problem that he developed some tools to help deal with it.

One of the tools he developed is the “Spectral Edit Multi-Tool” (Audacity Manual). When used with a “Spectral Selection” (a time and frequency selection Audacity Manual) in which both the upper and lower frequencies are selected, the multi-tool acts as a notch filter, but also the effect fades in at the start of the selection and fades out at the end of the selection.

In the track spectrogram view, mouth ticks are often visible as bright specks. To use the multi-tool, Spectral editing must be enabled and you must be viewing the track as a spectrogram, then you can select the bright speck and apply the multi-tool. The speck will then become less bright or disappear altogether. A keyboard shortcut may be set to activate this effect (Audacity Manual).

There is another tool that he made called “De-clicker for speech”. It takes a more automated approach to the task. I don’t know how well it works as I’ve never used it, but you can find it here: Updated De-Clicker and new De-esser for speech If you have questions about that effect, please ask in that topic.

Another thing that confuses me is that ACX recommends doing such things as low-pass filter, limiter, amplify/normalize after making edits, not before.

Well, you’re welcome … I don’t know how my jumbled musings and questions could possibly help anyone, but I’m glad they did. :smiley:

How on earth? lol It feels like I’m just floundering around with basically no clue. At any rate, thank you.

Simple :slight_smile:

If that’s what ACX suggests, that’s what I do. They also say, down in the fine print… somewhere, to avoid over-working your track. As I say elsewhere, I have refined it down to either Normalise, and maybe Limit. If it needs it.

But yes, after edits always. Take out silences, long breaks, obvious pauses. These will affect your RMS readings. Which are an average, right. Think on it. 1111222345611111. What’s the average? Try 123456. What’s the average?
Example 1 is your unedited track. Example 2is edited. The RMS is your average.

Save yourself heartache. If ACX say ‘do it this way’ …


Thank you for responding so quickly after my panic attack, Dana. I was too brain-weary to reply last night. I look forward to studying the video you mentioned.

In answer to your questions:

  1. I believe my gain is all the way up on the Solo and recording volume is all the way up in Audacity. I’d have to double-check to be 100% certain, but IIRC, that’s where it’s at.

  2. I’m about 4 closed fingers away from the pop filter, and the filter might be about … 2 inches or less away from the mic. I haven’t measured exactly, but I’m very close to the mic when I speak. Closer than I was when I first started posting questions here (when I was a full open hand away from the mic); keep in mind, though, my hands are small so my ‘measurements’ won’t necessarily be the same as others. If I move any closer than I already am, my voice becomes overpowering/boomy so I don’t think I can - or should - move any more than where I already am.

  3. Isn’t max noise floor that number I’m given when I run ACX check? If so, it’s pretty much always in the -70s.

    But, like I mentioned in a previous post today, I THINK I might have stumbled on the problem. At least I really, REALLY hope I have. From what I can discern, the long pauses that I had in some areas were contributing to the strange deletion of portions of my track. When I went back and removed a lot of the extra space that was there, the problem seemed to go away. I’ve been throwing all kinds of effects at it since then to test my theory and every time I use the effects on the track (or tracks) where there were very long gaps in between talking the problem shows up BUT when I remove the gaps it goes away. I’ve done this on all the tracks merged together into one long one, as well as testing the tracks individually. Same thing each time (unless I’m somehow missing something, which is entirely possible - I think my brain is fried at this point).

Now the question that remains for me (at least in this particular moment) is how to resolve the difference between what ACX suggests as an order of operations, and what’s been suggested here. ACX says to do edits first, then master:

Record & Edit


Steve mentioned the importance of removing DC offset (or reducing low frequency noise - I guess that means running a low-pass filter?) BEFORE doing edits because it can cause clicks. If I don’t run the low-pass filter and just want to remove DC offset, I can do that by using the Normalize effect (with the ‘normalize maximum amplitude’ UNchecked), correct?

Also, I thought I read somewhere (either here or elsewhere) about the importance of a certain sequence when running effects because if it’s done out of order, you can ‘undo’ what you’ve previously done. But I can’t remember now what, exactly, that was talking about. Maybe it was here and I should go back and re-read the last few pages carefully. There’s so much to absorb, I know I’m missing stuff along the way as I try to process everything in my mind.

Thanks SO much for your patience with me. :blush:

Yes, I’ve definitely started to understand RMS a lot better than when I began this. And I’ve gotten to where I can often tell just by looking at a track before and even after normalizing whether it’ll pass ACX.

@Dana Tucker

Dana wrote:
OK, that is your first problem. You should never have to set your recording gear to the max level.

On the SOLO Mini, with a dynamic mic, the Gain knob has to be right up, otherwise the mic isn’t working at its best. Having it all the way up doesn’t effect the signal of the dynamic mic, which is "weak’ naturally. It does in fact allow the mic to work at the level that it’s designed to. It’s “natural” level. Controlling the gain on the SOLO is really only effective with a condenser mic, which uses the phantom power. The dynamic of course doesn’t use the phantom power.
If you were in fact to then put an inline gain amp in, between the mic and the SOLO, you would then overdrive the signal. bad signal.

Stand-alone mic-pre and converters such as this are chronically low on gain. I note that they provide, on average, 40dB of gain which is simply not enough for dynamic (moving coil) or unamplified microphones. I have never used my Shure X2U or Behringer UM2 at anything other than full up. I complained to Shure since an X2U is inadequate for use with their own popular SM-58 and SM-57 microphones. They wrote back, “That’s the way it is.”

McElroy’s video for ACX shows him adjusting his interface for appropriate volume and I wondered how he was getting away with that. The answer is his microphone is a high-end and hot Rode NT1a.


The NT1 is even better…
My next major purchase… at about £200…
The 1A is only slightly under this spec. Minimal.

The World’s Quietest Studio Condenser MicrophoneLarge 1 condenser capsule with gold-plated membraneCardioid polar patternInternal Rycote® Lyre® based capsule shock mounting systemUltra-low noise, self noise of only 4.5dB (A)State-of-the-art surface mount electronicsAvailable with included SMR shock mount (NT1-KIT)The NT1 is a revolutionary new 1 diaphragm condenser microphone from RØDE.Although the body of the new NT1 closely resembles the NT1-A, the microphone has been completely redesigned from the ground up, with the only shared component being the mesh grille.RØDE?s design engineers approached the NT1 as a marriage of innovation and tradition, starting with the capsule which is a completely new design. Codenamed the HF6, it is the perfect example of RØDE?s fusion of artistic design approaches and cutting-edge manufacturing techniques, and features a sound signature reminiscent of the famous microphones of old while at the same time exhibiting extremely low noise.It has been developed with a focus on detailed midrange response, coupled with silky smooth high frequencies, and warm, round, bass reproduction to make the NT1 an absolute standout in its class.In another world-first for RØDE, the transducer itself is suspended inside the microphone using Rycote?s industry-leading Lyre system, minimizing external vibrations at the capsule level. The capsule is then married to high-grade electronics that have been designed to provide the lowest noise level of any studio microphone available. The NT1 is an incredibly quiet microphone, measuring only 4.5dBA of self-noise.Its body is machined from 6061 aluminium and then nickel plated for resistance against corrosion. Finally it is coated in a durable, military-grade ceramic layer, using advanced electrostatic application techniques developed by RØDE to ensure an extremely hard wearing finish that is resistant to scratches or marks.The NT1 is supplied with the revolutionary new RØDE SMR shock mount…

Like the others said, dynamic mics have to be turned up all the way or it just won’t work right (especially for ACX stuff); at least my mic does for sure. I spent hours playing around with gain and volume and the only way I could consistently pass ACX (after things like normalize or compressor then normalize) was to have the gain and volume at max.

I’ve heard of the NT1 but ended up going with the AT2035 because it was (currently) within my price range. I already have the Blue Yeti but didn’t want to fight with that one on projects like this. Speaking of ‘quiet’ condenser mics, I also heard about this one - the CAD E100s. But at more than $400, there’s no way right now. :stuck_out_tongue:

Yes, I’ve decided that condensor mics are not ideal for this sort of work, and my budget. When sales come in, maybe then a top end mic like a Rode. Meantime I’ll stick with my Shure55SH Series II. Which isn’t cheap by the way.
I’d like to try a ribbon mic which are also dynamics, as I believe they give a richer tone than a lot of other mics.
Probably look around for a vintage type.
I’m also looking around for a small/good valve preamp. ( vacume tube ) as I figure they will be able to gate out the mouth noises, that fast digital lets through. Just an idea, and finding such a thing may be an issue?

Yes, I’ve decided that condensor mics are not ideal for this sort of work, and my budget. When sales come in, maybe then a top end mic like a Rode. Meantime I’ll stick with my Shure55SH Series II. Which isn’t cheap by the way.
I’d like to try a ribbon mic which are also dynamics, as I believe they give a richer tone than a lot of other mics.
Probably look around for a vintage type.
I’m also looking around for a small/good valve preamp. ( vacume tube ) as I figure they will be able to gate out the mouth noises, that fast digital lets through. Just an idea, and finding such a thing may be an issue?
Now heres what I’m talking… Look for it on Amazon. Brilliant.

The original ART Tube MP put professional-caliber tube preamplification into the hands of thousands of musicians and recordists who wanted great tube tone.The new ART Tube MP Studio features ART’s ?OPL? Output Protection Limiter, which precisely and accurately controls and maintains the output peak signal.The OPL circuitry is crucial in protecting the next link in a signal chain - such as a hard-disk recording system or a sound card.The Tube MP Studio goes even further, with the inclusion of a VU meter for observation and measurement of the unit’s output level, which enables the user to keep his signal consistent with desired levels.The meter also reflects the impact of the OPL circuitry on the signal. For example, if the signal is ‘in the red’ on the meter, the meter will reflect the attenuation of the signal when the OPL is activated, and the signal is brought out of the ‘red’.The Tube MP Studio is the only Mic Pre in it’s class with sophisticated metering and OPL functionality. These features alone can add hundreds of dollars to the price of a processorThe Tube MP Studio can be used in a wide variety of applications including recording, project and home studios, where it’s metering functionality and OPL circuit really shines. It also functions as a direct box, with impedance matching and preamplification for line-level sources.The ART Tube MP Studio’s performance exceeds that of units costing many times its price… and it’s unique combination of features and flexibility will make it a must-have in any audio toolkit. Tube MP ? Studio Features * Provides Superior Preamplification for: Microphones, Instruments and Line Level Sources * Analog VU Meter * OPL? Output Protection Limiter * Hand-Selected 12AX7A Vacuum Tube * Variable I

How is that different from the Scarlett Solo? Will it work on the mics we already have?