Overthinking about troubles getting started with narrating

Special Interest Groups Audiobook Production
,
1

Hello! I’m very new to the world of using Audacity to record (literally never recorded myself like this before a month ago, no voice acting experience, etc). I jumped in to narrate some podfics, enjoyed the process, and got some nice compliments on my voice, which made me begin to look into the audiobook narrating scene. I’m having some problems, and am worried that I should stop throwing in money into something if it’s not feasible for me to turn this into a side gig. My major thing is that I always seem to be dealing with a lot of saliva, which turns into a lot of clicks and mouth noises. Using the Paul declicker seems to help, but I really am not sure if this is something that is gonna keep me from making the switch to professional. My research on this forum and elsewhere has left me confused on whether it’s alright to use declickers at all when submitting to ACX, because it sounds like they want little to no processing (or at least evidence of it). I keep switching between using it on the whole thing and piecing it in, because it likes to soften my ending consonants as well at times, but it’s a lot more work to piece it in.

For my other issue, let me explain my current setup.

Using Audacity 3.7.1. I switched from using my phone to record to using a cheap $40 USB mic from Amazon (and now have a Rode NT1/Focusrite solo coming in the mail). I’m set up in our upper room above the garage which actually has some sound sealing in the door, so I hadn’t really bothered with much else to reduce noise. The downstairs heater sometimes kicks in (upstairs I turn off), and airplanes and cars sometimes come by, but even with those I actually can get between -55 to -65 db in noise regularly, so just a tad bit of noise reduction seems to do it. Using my own editing process (high pass filter, filter curve, normalize, limiter) usually results in passing everything but the RMS (I’m on the quiet side), which seems to fine for podfic, but not if I’m trying to consider ACX. Whenever I try and use the audiobook mastering macro (I believe the latest one I’ve found was September 2024, 36Audiobook-Mastering.Macro), it always seems to really blow up my audio and then cause all these flat tops with the limiter. I’m wondering if this is because I’m not super great at keeping my loudness consistent, which is throwing off the Loudness Normalization step.

The last thing is I just went and bought the materials for the kitchen table studio noise setup, put it together on my desk, threw up another moving blanket behind me on a laundry line, recorded (mic on the mic stand clamped on the front of the desk and kind of angled into the opening)…and somehow ended up with basically no noise change if not worse (whether due to my room already doing decently enough with no echo or…?). I know I have the nicer mic on the way, but I have a feeling that’s only going to make background noise and mouth noises worse, so I’m feeling a little dejected right now and (as my name suggests), overthinking if I should just return all the materials. Also wondering if I should maybe have gotten a dynamic mic instead of condenser, with my mouth issues.

Here’s a sample of the room without the noise studio setup (yes the heater was running in both of these).

And here’s one with the blanket setup. Which, actually ended up with more noise, funnily enough.

Thanks all, this seems like a great little forum, I’ve read through a lot of posts while troubleshooting stuff, but I finally decided to just post instead of endlessly overthinking everything.

2

The very worst thing that ACX did during the pandemic is kill off Voice Sample Evaluation. It used to be possible to submit a minute or two of voice sample and a human would evaluate you and post back. When I did it, they loved the quality of my file and data, but they didn’t much like my mouth noises. I didn’t give up the day job.

One might say that ACX does offer an evaluation in the form of ACX AudioLab.

https://www.acx.com/audiolab

It turns out that they analyze the same things that Audacity ACX-Check does, but we analyze noise. They don’t.

I need to go back and read through that. You packed a week’s worth of experiences in two paragraphs.

Not everybody can sell milk, Catskill cows or not, but it might be handy to post some of your target work. The goal is sounding like you’re telling someone a good story over cups of tea.

If the work is mono, you can go out to 20 seconds. The forum will cut you off if the work is too long (2MB).

Koz

3

I wrote a dead-simple process for reading a chapter.

======

Read your chapter with wired headphones in a quiet, echo-free room. Repeat whole sentences if you make a mistake.

Stop.

File > Export the work, errors and all, as a WAV (Microsoft) file for protection.

Edit the work to get rid of word-o’s, stumbles, stutters, tongue ticks, lip smacks, and other errors.

Tools > Apply Macro > 36Audiobook-Mastering-Macro.

ACX-Check. If it passes and you like the way it sounds, Export a WAV (Microsoft) file as Edit Master and go on to the next chapter.

====

Most “How To Read” tutorials go on for pages and pages. This is what happens if everything goes OK. As you discovered, you may need a noise reduction boost at the end, but that should be about it.

There is one caution. I’m not using Audacity 3.7.1. I need to install that and see what the differences are.

Koz

4

Thank you! I’ll check out the audio lab. I noticed you mentioned noise reduction at the end; I had been doing it at the beginning, so I’m wondering if that makes a difference.

I’m curious what you think about the noise in the two clips I had in my post if you get a chance. I thought it was strange that the blanket studio setup seemed to have a higher noise level, granted the clips were taken on different days. I did just read a chapter in it again and my average noise level came out better at -65. But I’m starting to wonder if I even need that setup if I’m getting about the same noise levels on average without it. Perhaps I should wait for the new mic to get in and test that again.

5

There is one theatrical note. One forum poster started out sounding weak and frightened. They may have had other problems, but that one stood out. Over the course of postings, they got better and better and after we got done solving other problems, sounded pretty good.

One of my favorite audiobook presenters is Sarah Vowell. Nobody will be hiring her to announce television network news, but I have all her CDs.

Koz

6

Your background noise seems to be gentle spring rain in the trees “electronic” noise. FFFF. However the voice volume in the two clips is different. If you boost the weaker of the two voices, you also boost the background noise.

Noise is a little magic. I use a different standard than ACX. They like -60dB as a noise limit, but it’s not the worst idea to use quieter -65dB. That’s more likely to survive processing and production manipulation.

Also that -65 dB number assumes the type of noise you have. Gentle wishshsh sounds. There is a bad USB connection noise we call screaming mosquitoes. It has the characteristics of baby screaming on a jet. There’s actually special software for that one.

I would do noise reduction last. You can sometimes use Noise Reduction to clean up some other errors.

Koz

7

Your phone is not the worst way to record. This is my iPhone SE. That cable is for battery. The fancy-pants name for that is “Pressure Zone.” No you can’t make noises on the table or be in a noisy house. Some theater microphones work this way.

Yes, that is painter’s tape.

CatskillFarmsStudio-2_1778
CatskillFarmsStudio-2_17781000×791 232 KB

My voice passes ACX after editing with no further work. I added very gentle Noise Reduction just because I’m obsessive.

Koz

8

That’s Decaf Americano, no cream or sugar, if anybody asks.

Koz

9

We haven’t hit this yet, but how are you announcing? It’s not the worst idea to use oblique positioining (B).

That can increase the voice volume and can make the noise better. Since most wind noises go straight in front, it’s less likely to pick up mouth noises.

If you do that, it’s highly recommended that you wear the big fuzzy headphones so you can listen to yourself in real time.

Koz

10

I didn’t think about the voice volume being different, that makes sense! I do think I speak too quietly, so that’ll be something I need to work on. I am using the not straight on angled approach, though with the noise studio setup it’s a little scrunched in there, so I’ll have to keep an eye on that. I have my laptop under there too, because I don’t really have a way to do the whole punch and roll without it nearby (conveniently, at least. Without lugging monitors back and forth). I did try once to put it outside and do the test, I just didn’t notice much difference in the noise levels.

That’s really interesting on the iphone! I was finding that my volume was even lower with mine, though I was speaking into it about 5-6 inches away. Anxiously awaiting the new mic to come in, though looks like it’ll be later tonight.

I’m considering doing the ten day free trial with iZotope, just to see if their subscription-level mouth declicker works on my voice better than what I’m using. I honestly think I’m just perpetually dehydrated, something I need to fix as well :sweat_smile:

11

I don’t think we did this step.

Announce some of your own work—work that you’re comfortable with.

You can go out to 22 seconds in mono WAV. Leave one or two seconds of hold-your-breath silence at the beginning.

Cut it to sound like you want it to, but don’t fix the mouth noises. Post a sample that’s too noisy in your opinion, but don’t tell us where the noises are.

Koz

12

Sorry for the delay, I received my new microphone in the evening, only to realize I couldn’t even use it until I got a headphone adapter to fit into the interface. So it took me another day to out and grab that.

I can still go back for a sample, but now I think I’m trying to figure out this whole condenser microphone vs my old usb, trying to understand the gains, etc.

One thing I noticed while messing with it last night, is that with my usb microphone, I would end up with some waves that would jump up above the rest, sometime approaching 1 on the y-axis, but never really ran into clipping issues. While using the new microphone, I’m noticing it seems to be clipping/distorting before the waves even reach much past .5. As I’m listening back to my practice chapter, I’m noticing more of that fuzziness/slight distortion at sound levels that seem below where it was on the usb mic. I don’t know if this is just me not understanding the interface settings, or just needing to get used to the sensitivity. Not sure if you happen to have any advice on the gain, otherwise don’t mind me as I muse through my practicing.

I did end up with more noise, but nothing that didn’t seem to be fixed with the mastering then a little noise reduction.

I do love being able to listen to my voice at the same time - never realized how useful that would be at detecting when I’m too loud/quiet.

I’ll record another sample when I figure out my settings.

13

Oooh, actually I just discovered it might because I’m recording in mono with a stereo interface. I’ll try recording in stereo and delete the blank track next when I try again tonight. Baby steps!

14

That’s a standard error. “My voice is clipping at 50%!!” There can be other problems, but by far the most normal one is a conversion between stereo and mono.

Condenser microphones require less work for your voice to create a sound signal. That’s why they appear in studios so often.

This “studio accuracy” can make mouth noises much crisper and easier to hear.

They’re much easier to overload by accident and so don’t very often appear in News Gathering. Don’t blow into one.

They have electronics up at the head. That means you either have to install batteries, use power from the USB cable, or have “Phantom Power (48v)” coming up the cable from the mixer, recorder, or interface. That’s what that number means.

Koz

15

Finally back! I’m still experimenting with the new mic, my noise is definitely worse, which makes sense. But I’m actually able to get my stuff into the correct specs with mostly only the mastering macro, which seems good.

Even without my heater on, the noise seems pretty up there.

I’ve been trying out the Izotope free trial, and the mouth declicker actually seems to be working pretty great for my mouth noises, though I am noticing some of my ending consonants are getting picked up and softened by the plugin. I’m finding myself taking extra time replacing the missed sounds with the non-declicked copy, but it’s taking a little more time than I’d hoped. Perhaps I shouldn’t bother with that.

I do need to get a longer sample out here, I’ll work on that sometime too, though the dreaded mouth noises seem to coming through clear enough in that sample below:

(Side note, I’m not positive I am saving these files right, as far as going from stereo to mono. Listening to this sample on audacity comes through in both ears, but when I click the preview, I’m only hearing it in one. Curious. I usually right click to split to mono, then delete the second empty track.)

16

Again for clarity. Mastering applies a rumble filter first to get rid of low-pitched competition (busses, thunder), then RMS (Loudness) followed by gentle tip and peak adjustment. You should pass RMS and Peak no matter what you throw in there. That leaves noise which is always the most entertaining problem.

A note that Mastering works best in mono (one blue wave). I did that because ACX really likes mono and designing mono corrections took enough time.

The goal for this sample is to master it to about the right loudnesses, export to WAV and post it to the forum.

If that’s what you did this time, then Something Is Broken. First, the sample is in stereo (two waves)

Screen Shot 2025-01-11 at 5.19.10 PM
Screen Shot 2025-01-11 at 5.19.10 PM633×439 19 KB

Left on top. So I’m hearing it in my left ear.

[time passes]

I got it to sound reasonably good and pass ACX-Check.

Screen Shot 2025-01-11 at 5.42.13 PM
Screen Shot 2025-01-11 at 5.42.13 PM423×186 11.8 KB

But I had a pile of changes to make to get it there.

I converted it to mono and mastered it.

I need to go back to the edit and write down everything I did. Please note the first two seconds are almost completely silent. I did not apply noise reduction. That is your actual studio sound. I identified the offending sounds and noises and dealt with them individually.

Also note you don’t sound harsh any more. That turned out to be a squeaky, scratchy, high pitch boost coming from … somewhere. That was a separate filter I applied.

And lest I forget, your presentation and delivery is much improved. You sound relaxed and not terrified.

I have one caution. Since you are using a multi-job computer to capture and process your voice, once you get it settled, you shouldn’t use it for anything else until you get your book done. ACX requires all your chapters to match and there is no easy way to force other applications and software to leave the settings alone.

Isn’t this fun?

Koz

17

Missed one. You’re on Windows, right? If you’re doing the Windows Updates manually, stop until you get the book finished. If you’re letting Microsoft do the updates automatically in the background, there are ways to make it stop—for the duration.

Koz

18

Oh my bad, I sent the raw audio in without mastering. Throwing in mono and then just the macro got me within passing here below. Didn’t do anything else (besides rerunning the loudness normalization to -20, since I realized I had slightly adjusted the macro to -22 for reasons that I keep meaning to change back now that I’m getting a bit more leveled audio).

Curious about that high pitched boost. There’s something electronic-sounding to that clip you managed, but I don’t have trained ears at all, so it might be something with my headphones.

I am indeed using Windows, so those are good notes!

Eep about the laptop though. I work from home on it, so not sure I can get away with not using it for anything but a book. But I’ll keep that in mind. I found a monitor and threw it into my little area, to keep my laptop out (but did I do that for that sample…well shoot now I can’t recall. I think I did).

Thank you so much!!

19

We are warned against messing with the Mastering values and settings. The tools are balanced and depend on each other. For example, did you see the LUFSLevel setting while you were changing the RMSLevel? Those two settings have to match within their own worlds to sound right.

I’m sure you noticed that the mid-point between the two ACX RMS limits (-18 and -23) is not 20. It’s 20.5. That difference is used to make up the volume difference sometimes encountered in the Peak Limiter.

Mastering doesn’t have any active audio processing. In general, this means that the performance after Mastering still sounds exactly like you.

There’s another much less obvious Mastering trick. You can apply Mastering more than once and if the performance doesn’t need correcting, nothing happens.

Etc. There’s a lot going on in those three programming lines. (The top line is info and comment).

And that last clip you posted sounds too sharp and harsh. I would have a harder time listening to a story in that voice than the one I posted.

One of the filters I applied to “clean up” the sound was a hum filter. You have wall power socket noise in your file (120Hz if you’re counting). Sometimes removing that with the pre-baked tools can leave a slightly metallic sound. I know what the pitch of the interference is, I can try to get rid of it using more gentle methods.

That will need to wait for the Starbucks Infusion in the morning.

It’s good to note that a forum poster a while ago wanted to read for hire from his apartment in Hollywood (the real place). He had a terrific voice. We did get him working successfully after about a year of stomping out problems similar to yours.

He’s the record.

Koz

20

Thank you for the tips on mastering! I went back and fixed the rms level I changed in the macro, I’ll be sure to leave that alone.

I went and found one of your old threads on the decrisper plugins you created to see about dealing with that harshness, but I can’t seem to get my Audacity to recognize them (perhaps I need to do something with the xml file to make it readable). In the meantime, I used Paul’s de-esser plugin instead on some of my recent stuff, and that seems to help just a bit, though I’m not quite sure if it’s enough.

I did have my daughter’s hoverboard charger plugged into the outlet below my desk, so perhaps that wasn’t helping!

That’s great to know, here’s hoping I can manage something in perhaps a little less time! It’s exciting to see the jumps in quality going from zero to here in a month, so I’m looking forward to continuing to tweak things as I go.