At2020 & audiobox

My friend who is a sound guy says that my background noise is a signal to noise ratio .

Here’s a wav sample made by audition

Do I need to use audacity to get help? I’ve used both in the past. Everyone I’ve asked says they just plug their mic in and they have no background noise, I don’t get why I’m so blessed to have it. The my sound guy friend says there is always some static noise and fan noise. So how in the world do I get there to be no background noise like I hear on youtube videos with people talking from their home or streaming without a studio? When I go to remove noise like the 5 second length of no talking taking that noise print and then applying it to the audio there is still background noise when I talk.

How do I fix this so called signal to noise ratio?

Everybody has noise. The act of amplifying your microphone signal loud enough to be useful creates noise. That’s usually the rain-in-the-trees, ffffffffffff.

If you lead an impure life, you could have a USB microphone making computer noises also known as the “Yeti Curse.” This is in addition to #1 above.

Turn up the volume when I stop talking.

Then if you really ticked off a significant deity somewhere, you could also have computer fan noises, dogs barking, trucks driving by, etc. etc. I’m not at a machine where I can listen, but any one of those or any combination is enough to make you fail ACX AudioBook compliance. Is that the goal? People tend to go straight into the technical weeds and not tell us what the goal is. That can make a difference.

The reason the ratio is important is in the instance where you have a microphone and recording system that is truly terrific and a wonder to behold, but your voice is weak and very quiet. So you boost your voice volume high enough to be useful and the noise volume rises, too. The comparison between the two is important.

So fill in the holes. If you intend to read for AudioBooks, we have a number of tools and filters that can help. I’ll listen when I get there.


I tried a Virtual Audio Cable combined with noise reduction in Audition and these seems to be pretty much where I want my audio to be. It’s not exactly perfect, there just isn’t any real background noise except the ghostly like whispers at times.

I think this is what a lot do to get their audio to a perfect quality. Honestly I don’t really know why audition does a better job at noise reduction in real time then applying it to an entire clip. Logically I guess it makes sense since there’s
less every time for it to measure, just seems odd.

So all in all is there any edits I can do here to make it even better, it’s pretty close to that perfect area I want.

The only issues it’s doing I can tell is, the slight not picking up my voice 100% which is obviously caused from the noise reduction, it doesn’t do this much at all as an issue though it’s still noticeable. Also possibly get rid of the ghost like whispers, so maybe there is like 1 more effect processing that fixes these 2 things?

Any other tips on getting this as perfect as studio quality as possible, this is for Youtube video and Streaming on Twitch, basically it’s for human ears.

Are either of those clips “raw?” No processing? Straight capture and post the sound file?

Without that, we’re just commenting how well you read, not sound quality.

If you have odd artifacts in your Noise Removal, then you may be using the wrong Audacity. Noise Removal in Audacity before 2.1.0 had many artifacts and problems. Noise Reduction starting in Audacity 2.1.0 much less so. A very common problem is overprocessing. It’s a tail-chasing thing. OK, now how to I get rid of the whisps without breaking anything else?

We’ll tell you specific details when we hear the raw clip.


I’ve attached a raw recording in Audacity with the AT2020 and amp.

Thank you for your posting. I’ll give a listen on The Big System when it’s a more reasonable hour. It’s not unusual for us to be able to create smoother works with less processing or, in some cases, custom tools.

Are you doing all this on Audacity 2.1.0 or 2.1.1?


2.1.1 windows 8.1

I think he’s right about the fan noise. If you look at the spectrogram-view of your “raw” take, the noise consists of narrow horizontal bands, the worst is at about 2000Hz …
''raw'' take , waveform and spectrogram views.gif
That’s consistent with a motor , like a computer fan.

They use a noise-gate : that squelches the the hiss-noise to silence [only] when no-one is speaking.

You can download a noise gate for Audacity here …

NB: noise-gates don’t reduce the hiss which occurs during speech.

Tip: apply this “No Infrasound” equalization before applying a noise-gate …

Yes, but it’s doable. Attached raw-patched.wav. I don’t hear any artifacts and the background noise is almost non-existent.

I know you don’t need this, but see second attach, last sentence. You could read an audiobook.

I need to drift away for a little bit. I’ll post how I did it. It’s strictly post-production. Audacity will not filter anything in real time.

Screen Shot 2015-07-19 at 10.14.38.png

Download and unzip this custom filter. It’s a rumble/earthquake/thunder filter. Many field sound shoot mixers have it built right into the hardware. Steve designed this non-hardware version. You have to install it in the Audacity Equalizer tool, so that part is a little fiddly. Yell if I lose you anywhere in here. These are mostly pre-packaged instructions. I got tired of writing them over and over.

Adding Equalizer Curves
– Select something on the timeline.
– Effect > Equalization > Save/Manage Curves > Import
– Point at LF_rolloff_for_speech.xml > OK. (it won’t open the ZIP. You have to decompress it)
– LF Rolloff for speech now appears in the equalization curve list.

Open the sound file. Select the whole thing by clicking just above MUTE.
– Effect > Equalization: LF Rolloff for speech, 8191 Length > OK

You may not see anything happen, but all the truck-driving-by rumble is now gone. Those low tones can screw up the other tools and filters. Don’t forget to scooch the Length slider all the way up.

Drag-select your pure Room Tone (I picked 0.5 to 1.5 seconds)
– Effect > Noise Reduction: Profile.

Select the whole thing by clicking just above MUTE.
– Effect > Noise Reduction: 18, 6, 6 > OK.

Select the whole thing by clicking just above MUTE.
– Effect > Normalize: [X]Remove DC, [X]Normalize to -3.2 > OK

That particular Normalize is designed for AudioBooks. If you need it as Loud As Possible, change the -3.2 to 0. Adjust as needed.