Unwanted "squelching" on live recording


I’m using Audacity 2.3.0, though this started while using 2.2.2; I updated and reset all preferences in hopes that the problem would go away, but it hasn’t. I’m on Windows 10.

As of Wednesday, out of the blue, I have begun experiencing an effect something like squelch or noise gate during live recording. This is not an effect I am applying to existing recording. It appears to be happening as I record, and I would like it to stop.

When I play back the brand new, unedited track, you can hear background noise in the silence before I speak and while I am speaking. But during the first pause in speaking, the silence squelches down completely. Ditto in every pause thereafter.

It sounds very unnatural and distracting, and, like I said, it just started happening out of the blue without me knowingly changing my settings in any way. (I don’t think there was even a Windows update to blame it on.) I tried using that initial silence with background noise as a sample for the Noise Reduction filter, but it can’t remove enough of the noise from spoken sections to make the difference between speaking and not-speaking less stark.

To isolate the problem, I have tested this with another sound recorder (Free Sound Recorder); it doesn’t happen there. So I’m pretty sure it really is audacity doing this.

If anyone knows how to get this unwanted effect to stop, or can suggest why it started, I’d be grateful.

Audacity doesn’t (can’t) apply effects during recording or playback.
Check that all Windows Sound “Enhancements” are disabled. See: https://manual.audacityteam.org/man/faq_recording_troubleshooting.html#enhancements

I have - using a Blue Ice Snowball, it has no Enhancements in its Windows properties; and again, I’m not getting the effect when I record with the same microphone into Free Sound Recorder.

I know it sounds impossible, since Audacity can’t apply filters live, but something is happening to produce an effect that sure sounds a lot like squelching on a live recording. Shall I upload a sample? I could try recording two files, one through Audacity and one through FSR, so you could see the difference. I may also be able to find a recording on Audacity before this started happening, for comparison. Let me know what format is best to use for this purpose - export to mp3, or just the raw aup, etc.

Sure, that may help. Short samples (around 6 to 10 seconds) in WAV format may be attached directly to your post (see: https://forum.audacityteam.org/t/how-to-post-an-audio-sample/29851/1)

Thanks. I’ll try to get some samples up tonight or tomorrow morning.

Do you use Skype and is it running back there?

Do you like to record Internet music or sound?

Audacity can apply one tool in real time. Edit > Preferences > Recording > [X] Sound Activated Recording. But that doesn’t do what you got. That stops recording dead if it doesn’t receive any sound. It doesn’t produce a blank track of the right duration.

background noise in the silence before I speak and while I am speaking

OK, so you’re speaking. Into what? In detail. Part numbers.


Being an obsessive engineer, I would be "OO"ing into the microphone to see if I could sort which volume caused the trigger.


That would be much more valuable than reading a shaving commercial and leaving it up to us to figure out where the triggers are.

If you do it slowly, there may be Something Funny happening at the trigger point that would give us a clue.

If the test is in Mono (one blue wave) you can post a 20 second WAV file on the forum.


One more. Does the Audacity bouncing light sound meter drop to the left when the “squelch” kicks in? It’s possible not. If it doesn’t, that’s a terrific clue.


All consistent with Windows enhancements, (a Windows-update can enable them).
You should disable all the enhancements in recording & playback tabs …

Hi y’all,

So, sorry it took me forever to make those samples. Turns out the one recorded on audacity from before this started happening, I’d already processed the heck out of it, so it wouldn’t be useful. So I made two samples this morning, one with Audacity and one with Microsoft Voice Recorder (had to remove FSR; it stopped working after scrubbing for spyware). The latter is at a lower volume than the other, but I think the difference is telling. I just recorded the intro to the voice recording I’m supposed to do this morning for a local volunteer org, but I can redo with something more standard if you need me to.

As for the rest of the questions:

I’m recording speech into a Blue Snowball iCE microphone for both samples. I can’t find a parts number on the thing, but it’s this: https://www.newegg.com/Product/Product.aspx?Item=0H9-00C9-00001

Regarding enhancements in Windows Sound, I can’t imagine how those system-wide settings would affect recording and playback in Audacity but not in MS Voice Recorder? Nevertheless:

In Recording, there is no Enhancements tab for this microphone. In Advanced, default format is set to 1 channel/16 bit/44100 Hz and both Exclusive Mode boxes are selected.

In Playback, Realtek Speakers/Headphones also has no Enhancements tab. In Advanced, default format is 24 bit/48000 HZ. Exclusive Mode boxes are unchecked. Spatial Sound is off.

Audacity’s Sound Activated Recording is not enabled.

Sound meter: I expanded that sucker to the width of the whole screen so I could get some details, but this is still going to be kind of vague. When I begin recording at the beginning of the track, and I let it just record background noise (mostly the hum of the computer and the house heater) the meter bounces around the -48 to -45 area for about five seconds and then drops down to -57 and below. Then I say a word or two and it jumps up high. When I stop talking it, comes down to absolutely nothing before flickering, just barely, so low down it hasn’t hit any numbers on the scale; and it does that for the rest of the recording. (I can upload a quick screen capture video later.)

I don’t use Skype. No other recording programs that I know of are running in the background.

I will try the OOOoooo…oooooOOOOO test later, but I’ve run out of time this morning unfortunately.

Thank you for your help!

I can hear what you mean, and also observe a huge difference in the spectrogram view. Something is process the heck out the Audacity recording.
Assuming that the recording from Audacity is just a straight recording and you have only recorded and then exported as WAV, then you need to look at what else is running on your computer. Open the “Task Manager”, and identify everything that is listed (Google can be a big help with this).

To start the Task Manager, press Alt+Ctrl+Del, then select “Start Task Manager”.


I expanded that sucker to the width of the whole screen so I could get some details

Mine lives that way. This probably isn’t going to be visible…

In that configuration, I can set the meter to maximum range (96dB - in Preferences) and still have good visibility at the right where my voice is going to be. Note -6dB is perfectly clear and that’s where occasional voice peaks are expected to appear.

Can you tell the puzzled looks over multiple time zones?

There are two branches here. You can get stuck in the “If it walks like a duck” thing, where everybody fans out over the countryside with the howling dogs looking for “Squelch.” But it may be some other error that acts like squelch.

You are experiencing one of the reasons it’s sometimes best to stop using the computer to record sound.


Would you like the conspiracy version? There is “background processing” on your machine to make the Black Ops or Government Surveillance better able to pick up your voice and gain incriminating evidence.

Make sure you keep a radio or other music going in the background if you discuss sensitive topics. That’s still not a bad way to cover up your voice.

And speaking of Black Ops. Does it keep doing that if you disconnect your network?

This is Win10, right? Have you tried a clean shutdown? Shift-Shutdown, wait, and then start. Regular restart and shutdown doesn’t clean out everything. Pay attention if it takes a silly long time to shut down.


It does sound like a duck. I can almost give you the part numbers for commercial devices which sound like that.


Blue Ice Snowball

What does the recording device say in the little window next to the microphone?

Screen Shot 2019-01-26 at 9.20.24.png
Disconnect the microphone wait a second and then Transport > Rescan… Did your device vanish?


Just had a thought, please save and post the “Audio device info” (see: https://manual.audacityteam.org/man/help_menu_diagnostics.html#audio_device_info)

Ritual reminder that I used the same microphone for both recordings, both the one that walks like a duck and the one that doesn’t–I would have thought that alone would eliminate the microphone from the suspect list. Ditto any number of system-wide settings, like the fact that it’s Windows 10 and particular sound device settings and what was running in the background – same things were running in the background when I made the “squelch-like” recording on audacity and the “non-squelch” one on Windows Voice Recorder. The only difference between those two recordings was the sound recording app.

That said, I’ll try to field the latest batch of questions, in reverse chronological order because that’s how VBulletin displays the thread in the Post Reply screen…

Here’s the Audio Device Info for the Snowball:

Device ID: 1
Device name: Microphone (Blue Snowball)
Host name: MME
Recording channels: 2
Playback channels: 0
Low Recording Latency: 0.09
Low Playback Latency: 0.09
High Recording Latency: 0.18
High Playback Latency: 0.18
Supported Rates:

It displays in the Recording Device drop-down menu as “Microphone (Blue Snowball)”.

Yes, after unplugging and then Rescanning Audio Devices, it does disappear from the list. (Ooh, and then when I plug it back in and rescan, it reappears. Cool feature! Up 'til now I’d been closing down and restarting audacity to get it to detect a newly plugged in device. Thanks!)

Occurs to me I haven’t tested out recording via other microphones into Audacity. This may be a “feature” of specifically the relationship between the mic and the program. Maybe? Is that possible? (Not that any of us would have thought this effect I’m getting is possible in the first place…)

“Assuming that the recording from Audacity is just a straight recording and you have only recorded and then exported as WAV”… Precisely the case, yes.

And, lest we forget, it only started happening when I sat down to do Wednesday’s (Jan 23) recording for AINC. This just started happening out of the blue, apropos of no windows updates or settings changes.

As for everything else that’s running on the computer - I’m going to have to get back to y’all on that. A deep dive on Task Manager will take a while, assuming we’re including Background Processes and Windows Processes, which I expect we’ll have to?). Besides which, this is not a factor that changed between the walks-like-it’s-squelched audacity recording and the fairly normal Windows Voice Recorder recording, so I’m kind of reluctant to spend a lot of cycles on that analysis just yet. But if y’all think it might be fruitful, I will do it!

In this post, I have underlined assumptions, not because I doubt you, but because “I” have no direct evidence that they are true (I can’t see your machine). It may be worth you double/triple checking any items that I’ve marked as assumptions.

Your previous test with Microsoft Voice Recorder was enough to rule out the microphone from the suspect list, assuming that both recordings really were made with the same microphone.

The “microphone drivers” are a different matter, as some audio device drivers provide “per application” settings. As far as I’m aware, the standard Blue Snowball iCE microphone uses drivers provided by Windows, so will have whatever abilities Microsoft decided to give in the most recent Windows update.

Sure, it’s not due to a system-wide setting, but Windows also provides some per-application settings.

That’s not the full settings. To get the full settings, click the “Save” button in the “Audio device info” dialog, then attach the saved file to your post.

So “something” must be processing the audio data from the microphone somewhere between the physical microphone, and “PortAudio” (the software library that Audacity uses to communicate with the computer sound system.

Assuming that you have an original and unmodified version of Audacity, there is no way that Audacity can modify the audio during recording. The ability to do so does not exist in Audacity. Over the years, there have been requests for this ability to be added to Audacity, but to date there have been no attempts to add this ability.

For completeness, there IS an experimental feature “EXPERIMENTAL_AUTOMATED_INPUT_LEVEL_ADJUSTMENT”, but as can be seen here, https://github.com/audacity/audacity/blob/master/src/Experimental.h this feature is disabled. This experimental feature was originally called “AUTOMATED_INPUT_LEVEL_ADJUSTMENT”. As can be seen by looking at the the code history, this has not been modified in the last 9 years, which was when Audacity was moved from SVN to Git (https://github.com/audacity/audacity/blame/d18553a3f0ca6797168c2c9bd28c1f61b8f063f8/src/Experimental.h). Even if this code was enabled (which it never has been), it would not account for the spectrogram changes that are clearly evident in your posted sample.

Yes, really useful :slight_smile: It would be nice if Audacity could automatically detect when devices were connected / disconnected, but we’ve not got that yet.

The “effect” is very well known, and is entirely consistent with Sound System “Enhancements” being enabled. The mystery here is that you assure us that there are no Sound System Enhancements enabled.

Assuming that there are no Sound System Enhancements enabled, then there must be something else running, between the microphone (hardware) and PortAudio that is producing the effect. We know that neither the microphone (hardware) or Audacity (genuine release version) can cause the problem, so it must be something between these two points.

so it must be something between these two points.

[Following this avidly.]

I would have thought that alone would eliminate the microphone from the suspect list.

Normally, yes. But this isn’t normal.

Do you use any other sound programs including Skype, Chat or Conference? Games?

I think it was Popular Electronics which had a “Tough Dogs” column. The writer had decades of experience and he would describe the unlikely causes for some of the symptoms presented. He also responded to write-ins.

Scene shifts to my formal electronics training. The instructor had an electronics design which saved the manufacturer a couple of pennies, but functionally unservicable if it ever broke. “Take that, whippersnappers!” No, we couldn’t fix it either.

I wrote in another post about following an internationally respected engineer around as he did a survey. He did everything I would have done, except he didn’t make any assumptions or pre-judge the result. I sit at his feet.