Audio file suddenly sounds like I got a squeaky toy...

Special Interest Groups Audiobook Production
1

I’ve been on an editing marathon this weekend for a book I have due this week.

I ran through first edits on this chapter, and then put it away for an hour of games to blow off some steam. Coming back to it, I started to notice an occasional odd squeaky sound, almost like a dog toy, under voice - it never occurs unless I’m speaking. I thought for a second perhaps something went wrong with a notch filter I’d used, and so I undid the filter. That didn’t fix it, so I exited the entire file, choosing not to save, so I could come back to what should have been fine - my first round of edits is only ever to clean up duplicate sentences and plosives.

Reopening the file, the squeaking sound is actually worse now. I thought it was my headset, so I chucked mine and grabbed a new set. Nope, still sounds like a shitty squeaky toy.

Any thoughts or ideas? I’m really loathe to junk the hour of recording and hour of edits I already have in it. Yes, I’m learning the hard way why you never ever begin edits without a copy of the original, untouched file. Damnit.

I attached a sample. Kind of hoping you’ll tell me its in my head and I just need to sleep.

2

Unfortunately its not a hallucination caused by sleep-deprivation, I can hear it too : it’s like clarinet, only when volume is loud.
I don’t think you did that in processing, it’s more like a physical resonance-effect on the original recording : the microphone / mic-stand / pop-shield vibrating at a particular frequency driven by the sibilant & plosive energy.

[ The pop-shield shouldn’t be attached to the microphone as it can transmit vibrations ].

You can edit out the clarinet using Audacity’s spectral selection tool, but each one has to removed individually …

It would be quicker to identify the problem in your set-up , then do a retake.
''just asking how I was' 'spectrogram, clarinet sounds highlighted.png

3

My pop shield doesn’t seem to be doing a damn anyhow; I plosive like nobodies business when I really get going. It’s actually attached to the desk, but it’s not always stable, so I wouldn’t be surprised in the end if that was what it was. I did a few dry runs this morning without a pop shield and I’m not hearing the squeaks. Which is fab, because I was worried it was a problem with my mic, and it’s only a couple months old.

You can edit out the clarinet using Audacity’s spectral selection tool, but each one has to removed individually …

I wouldn’t even really mind this, because it seems to be concentrated in areas throughout the file…but the sounds are pretty hard to identify in the spectrogram. Some I can spot immediately…others seem fully blended in and aren’t on the same frequency range, unfortunately. So it looks like I’ve got about 4 hours of rerecording to do. That’ll teach me to do large bits of recording without listening back closely. :angry:

4

You should still use a pop shield, if for no other reason than to protect the mic from spit accumulating on the diaphragm, which could permanently affect its performance.

If you can’t identify the problem with your existing pop shield, one can be improvised with a sieve and/or stocking material.

5

You should still use a pop shield, if for no other reason than to protect the mic from spit accumulating on the diaphragm, which could permanently affect its performance.

Of course! I have another lying around here somewhere that will (hopefully) be more stable and not cause me any weirdness. Thank you so much for your help - it would’ve taken me ages to figure out what was causing the noise!

6

BALLS.

Out of the two chapters I just rerecorded, the first was absolutely fine, which was what led me to believe it was my popfilter, which is still eating dirt on the floor right now. But the second, had the clarinet sounds scattered throughout again.

I’ve tried moving my setup, changing where things are, examining for loose pieces on the desk, and even just moved the whole shebang to my floor. Still squeaks. I wonder if its my microphone, then? Not sure why it would make the noises sometimes and not others, but I’m not sure what else it could be.

7

The only other explanation I can think of is sound leaking from headphones and being picked up by the mic, making a feedback loop, that would create noise which would also have a preferred [resonant] frequency, and could be restricted to loud sections. [ but I’ve never heard feedback sound like short bursts of clarinet : it usually rings for longer than that ].

That feedback explanation is only feasible if your voice is playing back live in headphones as you record.

8

I actually don’t tend to use headphones when I record. I did later this afternoon when I was trying to hear the sound as I recorded, but not during the initial recordings.

I just went and purchased a second mic, for the sake of experiment (and desperation). I grabbed another laptop, and went to town. Both have squeaks. Out of frustration, I even tried different software, and still had plenty of clarinet noises throughout.

I’m left thinking…is it somehow me? No, but really, I’m going crazy over here.

9

I grabbed another laptop, and went to town. Both have squeaks.

That would have been my guess. I don’t think it’s acoustic. Acoustic errors have a signature since they’re related to the sounds in the show. These don’t appear to.

So that leaves environment. Where’s your cellphone during all this? Do you have any BlueTooth equipment or connections? Do you have those Philips LED lightbulbs that change color with the time of day or other stimulus?

How are you reading your script? I saw somebody performing while constantly squinting at his cellphone and I wondered where his script was.

[Booming Kettle Drum Sound]

Oh. Right. I wonder if he remembered to turn off the bell.

Other than “headset,” we don’t know a lot about your setup. How are you recording, exactly?

I read through that several times. It gets worse while you’re editing? We should be careful about this, because if you use compressors or other strong-arm tactics in post production, you could be making your own voice louder, but also making the noise more prominent—and it’s always been there.


I’m just now casting my mind back and I have a live sound clip I made with a similar problem. “What’s that little beep in the dialog?”

Can you make it worse?

Make a 10 minute recording of nothing. Room Tone. You go make tea. Anything there?
Now make a recording where you present major sound groups. A couple of seconds of sssssss. several of shshshshshsh. MMMMMMM, ooooooooooo. etc. My problem seemed to get excited during sibilants in my delivery. I never went back to track it down.

Koz

10

The artifact seems to correlate with strong sibilance …
artifact seems to correlate with white-hot sibilance.gif
if that’s the case it rules-out interference from other electrical devices as the cause.

Any real-time processing hardware used before Audacity records the sound ?, e.g. compressor / expander / de-esser .
[ A de-esser would be triggered by strong sibilance ].

11

Yeah, but a de-esser’s reason for being is to suppress sounds, not make new ones. I need to recreate my shoot and see if I can reproduce the noise. We are walking around in the dark. This is not a common problem and I, at least, can’t think of a way to intentionally create the error. It doesn’t count if it violates Occam, either.

If the error can’t be controlled in any way, that’s not good news.

Koz

12

What about a mismatch in sampling-frequency between Audacity & the sound-card, so it’s unintentionally functioning as a frequency-divider , ( like those bat-detectors which make ultrasonic audible ).

It looks like the “clarinet” artifact frequency is ~1/10th the frequency of the sibilance corresponding with it …

If that was the case, the problem is occurring continuously, but only becomes conspicuous when loud high-frequency sound occurs. ( 1/10th of 200Hz is 20Hz which is inaudible , 1/10th of 7000Hz is 700Hz which is audible ).

If that is the problem the solution would be to make sure the sound-card & Audacity are operating at the same sample-rate.

13

So that leaves environment. Where’s your cellphone during all this? Do you have any BlueTooth equipment or connections? Do you have those Philips LED lightbulbs that change color with the time of day or other stimulus?

Cellphone stays in another room (partly for noise/interference potential, partly because I get all OOH, A NOTIFICATION). I literally unplugged everything in the house but the fridge when I first had my frenzy about this; I keep a small 40 watt bulb in a desk lamp, but even that is off right now.

I record with just a mic (yeti usually for warmth, sometimes an AT, tried both) and a laptop (tried a mac, tried two different dells), in a baffleboxed tent of sorts (see here: https://forum.audacityteam.org/t/i-passed-acxs-qc-but-the-end-result-sounds-horrible/42749/15), although I’ve moved it to a different wall since those pictures to avoid being on an outside wall. I own pre-amps and a few more technical doodads from years ago when I did more radio work, but for audiobooks, this has been really all I needed. I know the bane of audio recording issues is someone with a USB mic and a laptop, but for the most part, I haven’t had many problems besides a hum that turned out to be my radon (which is thoughtfully easy to turn off).

I never had this issue during previous recordings.

It gets worse while you’re editing?

I thought it did, but I’m pretty sure I was wrong. I think I just didn’t notice it until edits, when I was listening to things really closely. While I do normalize and occasionally a few other gizmos, I noticed it before processing - I was just doing basic edits - removing duplicate sentences, cleaning up occasional mouth smacks, etc. I did throw a notch filter at that particular file to remove a hum at 50.

When I record room tone, I get a whole lotta nothing. I talk, it happens, and I seem to notice it more with vowels. I’m now 95% sure it’s me. If I wear headphones and listen to it live as I record, I hear it happen sometimes. I apparently morphed into a $&^#ing squeaky toy overnight. I’ve tried a lot in the last two days - uber hydration, 8 hours of straight voice rest before recording, drinking frequently WHILE recording (water, not liquor, unfortunately, although that may change), and just listening to myself and repeating sentences in different ways until I don’t squeak, which is making each chapter take 3 times long to record and edit.

Which means I put you wonderful people through a lot of hoops!

I do have a question though - on your spectrogram, I can see them so clearly (beyond the fact you helpfully highlighted them) - when I look at the spectrogram myself, I can hardly ever find the buggers. Is there anything you’re doing that I could use to help me out? I’m going to TRY to move forward and aggravatingly repeat sentences until I don’t sound like a clarinet (until I can perhaps see a physician who will likely treat me like I’m nuts), but for the worst honks (and for the honks that occur during emotional scenes) I’m trying to figure out how best to diminish them without 5000 re-recordings.

Thank you guys so much for your help!

Edited to clarify: I can’t hear myself doing this without the headphones recording back live to me. I can’t feel a change in my voice or anything or that nature, so that’s why I’m not 100% sure its me. That, and sometimes it sounds oddly digital in nature. I’ve had my SO sit here and listen to me record while it happens and he swears up and down there is no underlying clarinet happening from me. I did a short recording in a different (but much noisier area) and didn’t have any squeaks, but I have also been able to get it to not happen by changing how I say a sentence. To end, I’m confused as balls, but I’m the only constant in each situation.

14

only becomes conspicuous when loud high-frequency sound occurs.

Yeah. That was mine. I got it immediately trailing a word ending with very aggressive SSS sound.

I gotta go back to my notes…

I just didn’t notice it until edits

The clip would have sailed right by me until it was flagged and noted.

I’m the only constant in each situation.

That’s not true. We’re confused, too.

But now I have something to chew on since I can semi-create a similar problem under speech conditions.

Koz

15

Which means I put you wonderful people through a lot of hoops!

It’s good to be able to say we know how to cure it when additional people arrive with this problem.

Koz

16

If all your tests were done in the same room (which has the tent) it could be something in the room/tent other than you.
Something capable of resonating like a bell, the shade on the desk lamp ? , or a tin/vase/jar ?
shade on desk lamp , tin with pens . glass jar.png

17

Like that?

Now I need to remember how I did that…

Koz

18

Correct me, but hasn’t this been problem at the leading and trailing edges of the sound? Those are traditionally magic with modulation effects. Maybe the system can’t handle the extra tones generated normally when a sibilant turns on and off.

Are you using 44100Hz for a sampling frequency? Does it change if you use video standard 48000Hz? Does your system support the Recording Studio sample rate of 96000?

Koz

19

I second that different sample rates be tried, [actually I proposed it …

https://forum.audacityteam.org/t/audio-file-suddenly-sounds-like-i-got-a-squeaky-toy/43159/12

20

Something capable of resonating like a bell, the shade on the desk lamp ? , or a tin/vase/jar ?

Ever since I moved the desk off the back wall last week, none of the tchotchkes are on it. I took the lamp off two -ish days ago when I was desperate to find the source of the issue and unplugged the whole house :smiley:

My actual setup is like this:

Except theres a missing baffle box behind the desk at the moment. I’ve tried with and without the laptop stand; I tried standing versus sitting in case it was the chair; and I do have an “announcing tunnel” thing (box with foam in it for the mic) but I hate recording in it because it well, sounds like I’m recording in a box. I tried recording with it last night, but the squeaks were the same.

Correct me, but hasn’t this been problem at the leading and trailing edges of the sound?

I feel like it happens more (when I’m hearing live) over vowel sounds. If I’m doing my “manly” voice, it doesn’t happen. I don’t have a high pitched voice anyway, but it seems to happen most when I’m talking at naturally. I initially thought that when I spoke very fast, it wasn’t occurring, but on listening back to some things, I find I’m wrong.

When I brought my mic out to my “main” desk, where I recorded my very first book that had a pervasive computer hum through the whole thing (slap my hands, I know), I felt it didn’t occur. But I only recorded like, 15 seconds worth of audio, so I’m going to try again tonight to read a few minutes worth and see if its something in the damn room.

Are you using 44100Hz for a sampling frequency? Does it change if you use video standard 48000Hz? Does your system support the Recording Studio sample rate of 96000?

Here’s my sample rate:

http://i.imgur.com/YNiCQtj.png

I know I can set audacity to say, 48000 or 96000, but I’m unsure by how I know if my system supports it, or how I ensure my sound card and Audacity are at the same rate? It didn’t happen in the last book, and it’s the same laptop and system, but god knows weirder things have happened. I haven’t tried yet to record at those samplings because unfortunately, I have to go to that damn “real job” thing in a moment, but I’ll attempt recordings tonight.