Getting rid of background noise

So… Regretfully where I live I live with 2 other people. One of them keeps the tv on in the living room up loud. So when the recording goes quiet you can barely hear the tv. It’s made worse when I use the Compression tool. I tried the “Reduce Background Noise” tool with no success. I selected a long section where I wasn’t talking to build a profile and then selected the entire recording and applied the tool with default settings, then the Compression Tool and it still boosted the quiet sections where you could just hear the tv in the background. I am thinking, and remember I am a newb at this who’s trying to make better quality recordings, that if I knew of a way to reduce any peak at or below a certain threshold to zero, where I talk would be unaffected but the tv sounds would be nuked. That or find a way to have Audacity would reduce the volume to zero, while recording, if the sound incoming drops below a certain point, then when it rises up above that level it brings the levels back up to where they were. So when I am talking it is recording but when I shut up it’s just recording a zero db signal. Or something like that. I don’t even have the terminology to describe what I mean. >.< I saw something like that in ODB but I use Fraps for recording the game and trying to use Audacity to record the commentary so I can adjust the levels, add compression, etc to boost my recording quality, like the pros do it. Also easier to snip out sections where if one of the two I live with pops their head in and starts yapping at me while I am recording, then the game audio continues just my commentary gets muted for a minute.

Oh, forgot again, this is Audacity 2.3.2


NB: a noise-gate will have no effect on noise which occurs when you are speaking.

Oh! Rewriting this post, read up on the Noise Gate a little. Ok, updated to Audacity 2.4.2. But I am running into a problem. It’s not applying the filter. I have it set to “Gate” “Link Stereo Tracks” “No” “0dB” “-12” “-29” “250” … It’ll look like it’s doing things until the last few seconds then it’ll stop and close the working window. If I check “Edit” the "Undo " is for the Import, not for the Noise Gate. It’s like the gate is either not applying or crashing. O.o

If you have the threshold set too high the gate will do nothing.

There is a “function” option in the noise-gate to “analyse [measure] noise level”,
(the threshold to use will be a little higher than that measured value).

Ok. It says -35 when I analyze a patch of the noise I am trying to get rid of. Using -35 as a threshold, select all, same problem, it isn’t doing anything.

Well… How about a few bad words here, that is if I am interpreting this correctly…

According to -

Noise Gate
(noisegate.ny) Download

Author: Steve Daulton.

Warning icon This effect requires the entire audio selection to be loaded into RAM. If there is insufficient available memory, the plug-in and Audacity will crash. The maximum length that can be processed is dependent on the sample rate, length of audio selection, operating system, and available RAM. Please test carefully before using this effect on long (> 45 minutes) tracks to verify the limits on your computer.

This is a 51 minute PCM Wav file. >.< So… I guess the question is, how do I know if the entire 550mb WAV is fully loaded? Length is 51 min, OS is Win 10, available memory is 33% of 16 gb, so about 6gb, and not sure on the sample rate. Audacity is set to 44.1 Khz 24bit.

You shouldn’t go back and edit old forum postings.
Nobody will see the corrections.
We work forward down the message list only.

You hit one of the sure-fire ways to kill an audio production (#4).

Screen Shot 2021-01-01 at 2.10.22 AM.png
Effect > Noise Reduction only works on sounds that never change such as microphone hiss or wall power hum.

Effect > Compressor brings the background sound and the performance closer together, so that’s the exact opposite of what you want. Even worse, Compressor can make it so the other tools, filters, and effects stop working.

Effect > Noise Gate is the answer for fixing this in post production. It’s not automatically adjusting and can be tricky to use. It depends on the main recording to be perfect, so the main voice can’t have problems of its own.

What’s your microphone? You might need a hyper-cardioid type instead of a regular directional microphone. Also a possibility is a shotgun microphone.

These are very directional and will ignore sounds coming in from the sides. Also very expensive.

You can also help a lot by making a mini studio. This is the Kitchen Table Sound Studio.

There was a video made by someone who got good sound isolation by just crawling under a heavy blanket while he was recording. You have to be good at not breathing to make that one work.

He did it to avoid room echoes, but that helps with environment sound, too.

Is there a goal? I don’t think we ever hit a goal in there. Are you producing a podcast? Audiobook?


It might be handy if you record a short test so we can hear what you’re doing.


Uh, yeah, I don’t control the TV, that’s my Uncle’s and he keeps it cranked. >.< As for editing old posts, I had said something stupid then later realized I had (learned more on my own and was like “Oh, yer an idiot.” to myself. Let me see…

@kozikowski - #1, recording in my bedroom and it’s not echoy. #2, did a test recording and adjusted the levels to only touch red not live in it. #3 I exported as PCM Wav and reimported after I updated Audacity. #4, well, as I said, I have no control over the tv in the living room. If you ever had to deal with an 80 + year old man, you know they are so set in their ways it would be easier to move a mountain than get them to change or admit they need hearing aids. >.< My Dad was the same way back when he was still alive.

As for my microphone, it’s a Purple Panda wired lapel mic. When I got the Blue Yeta Snowball I realized a desk stand wasn’t going to be ideal, I would need to invest in a bunch of stuff. A clamp scissor stand to clamp to the desk, a pop filter, and probably a shock mount as well. Just so I could pull it over closer. I sit about … * pulls over a tape measure * 2 feet … Umm… 0.6 meters? … from where the mic would sit on the desk and the sound levels were usually too quiet unlike head set or lapel mics. I settled on a lapel since I couldn’t find an open back headset and I found myself shouting if I used a closed back headset since I couldn’t hear myself talk. All except one Turtle Beach headset that fed the mic back into the ear cans in real time (something Windows can’t do, Windows always delays it and gets me royally confused). But the last of those I owned the Realtek control panel whined the impedance was super high. These Samson SR850s are nice.

Blanket idea might work, but not in summer, summer in Southern California is brutal enough as it is without adding a blanket to the mix.

Goals. To make my Let’s Plays a better quality. Normally I’d have Fraps record the mic as well, but that means I can’t really do anything to the game or commentary separately. Admittedly I live with my Mother and Uncle (Rent in Los Angeles is brutal, cost of living in L.A. is brutal, let’s be honest. >.< ) and they are another source of headaches. This last recording my Mother stopped on her way past to her room to ask me when I was going back to work (I am on a 2 week holiday atm) and later my Uncle stopped to tell me he tore his t shirt and made a joke about it. * eye rolls * In my normal way of doing things I’d have to clip those sections of audio out so, with commentary and game audio mixed, it means all sound would be muted for a minute or two to get rid of that. With it being separated I could clip just the commentary track and leave the game audio alone. Plus I can adjust and tweak the audio tracks separately.

So you want me to record a short test? Probably be easier than trying to stuff a 550 mb 51 min WAV file onto Dropbox, One Drive or something. >.< I’ll record something later so you can hear the tv and all that.

Back to the other question, if Noise Gate needs the entire clip loaded, how do I make sure it is fully loaded? The way Noise Gate behaves it’s obviously not since it stops short and closes without doing anything, even if I use the recommended Threshold.

So you want me to record a short test?

I do, yes. We’re not working on your problem. We’re working on your description of the problem. We’re also unsure of the quality of the voice and we’ll need that to tell you how to use the Noise Gate.


Ok, here is your sound test clip. Also I dug out my Snowball, though I can’t find it’s USB cable as of yet. >.< Old school USB printer cable to boot.

Old school USB printer cable to boot.

Also known as USB-A to USB-B.

You may not need it. You need to come up with a “studio.” Your job is not to peel off “background noise” from your performance. You need to separate two mixed performances from each other. This is a cousin to splitting the instruments from each other in a band or orchestra.

Audacity won’t do either one.

Even if you do get the gate to work perfectly, it’s going to give you TV-stained words over velvety blackness. The instant you both talk at the same time, you’re dead. See: 8 seconds in your sample.

Start looking for heavy blankets or furniture moving pads.

Do you drive? I know that’s a silly question in Los Angeles, but I have had people turn out very good voice tracks from the back of their car. You don’t have to worry about trudging through the snow to get there, either.


Noise-gate does more harm than good on that …

IMO buy uncle a belated Christmas-present of wireless headphones.

Christmas-present of wireless headphones.

Maybe not wireless. You can’t plug those into a microphone, interface, or sound mixer.


Well, so much for a cardioid microphone, which the Snowball is. It still picks up background noise, but only when I am talking. O.o

so much for a cardioid microphone, which the Snowball is.

It’s switchable with the little switch on the back. For directional, make sure it’s in position 1 and then make sure you’re talking into the company name. Cardioid microphones are not a gift from the angels. You still need a quiet room—or a blanket to hide under.

It still picks up background noise, but only when I am talking.

That’s the noise gate.

The only cure for your problem is to use a software generated voice and not try to record your real voice.


It is in switch position 1 and, no, unedited, unless VoiceMeeter (which I need as Win 10 doesn’t have a microphone boost for the Snowball, does for the lapel, but not the Snowball) is doing some sort of Noise Gate type effect. I barely understand it, followed a tutorial and when it’s quiet it sounds reasonable. Without VoiceMeeter it’s too quiet in position 1 and can be barely heard in position 2 so I assume 2 is the -10db pad setting. In position 1 I use these settings and it is good, doesn’t clip at all. But it does that noise gate thing on it’s own. I should have kept the test recording but I was disappointed. Started out good but then when I started talking is when I started to hear the tv again. >.<