I have an interesting situation. So I have this desk fan that runs and creates a little background noise. Not a big deal, but I need to sample it and remove it with several passes. In my macro for getting my audio edited, I have 4 noise removal passes at relatively low levels (Noise Reduction 4, Sensitivity 6, Frequency smoothing 5).
Currently I have to select the sample and then run my macro.
First Question:
Is there a way to put the selection automatically into the macro steps? For example, tell Audacity to select the first 5 seconds of the clip for noise (or another area) and then use that for the noise removal?
Second Question:
For increased precision, is there a way to repeat this? In other words, select first 5 seconds → run the NR based on that sample. Then again, select first 5 seconds → run the NR based on that sample. So that it reduces chances of artifacts and eating into audio as much as possible.
We do note that the Profile Selection sticks around until you close Audacity. Also, people have been known to save a short sound file that is the noise profile. There might be a way to program Noise Reduction to look for a special sound file rather than trying to guess at “the first three seconds.”
Skype, Zoom, and other programs look for long form sounds. Anything that hangs around for longer than XXX seconds must be noise and get rid of it. That also gets rid of music.
If you’re doing all this in a Macro, just repeat the macro code?
I don’t know that it’s written anywhere you can’t do that.
People have noted that repeating Noise Reduction even with the same Profile has different results than trying to do it all with super big reduction numbers.
So there’s no way to use the selection macros to tell it to select a portion of the audio and use that as the sample for noise reduction? The selection/sampling has to be done manually?
At least the first time. I think the Developers have been firmly resistant to programming Automatic Profile because it’s crazy more complicated than you think.
Note the effect is Noise Reduction, not Noise Removal. It used to be Noise Removal, but too many people were trying to Remove the Noise … to Zero. “Noise Removal leaves some noise behind!!!” I see the new Developers have returned to Noise Removal and Repair. We’ll see how that goes.
It is said from on high that you can make up for not having a studio (quiet, echo-free room) in software. We’ll see how that goes, too.
The thing is I live in Arizona and it’s hot as hell, and when I speak for extended periods I get really sweaty. I need something to keep me cool, and any desk fan is going to blow into the microphone to some degree. Haven’t figured out the best solution.
I did notice that fans don’t have low-low-low. They have High, Half and a Little Less Than Half. Anything slower than that and they have to special design the motors. Motors do not like slow. They work by juggling magnetic fields and the juggling act falls apart at slow speeds. You want to destroy a motor, stall it.
That’s conventional fans. You can get some computer fans to do some magic things. They don’t depend on velocity to do the traction.
There is a Live Steamers Model Train Club at the other end of Los Angeles. I was riding on my friends locomotive and we passed a woman and her train at a siding moving gracefully at no miles an hour. I said, “I want what she has.”
Just as another aside: I have found the Noise Reduction is pretty worthless on wide-spectrum sounds like hiss and wind noise. It works much better for pitched noise like 60Hz hum or the digital aliasing my DAC causes :^)
The issue is the air hits the microphone, so no matter what it will create some baseline noise. Not a big factor with a few noise passes, but was looking for a way to automate selecting then doing noise removal.
Roll off for speech doesn’t actually do anything. I tried some examples comparing a test recording with fan level 1 and 2, comparing between using roll off (no noise removal) and 3x or 4x noise passes. On fan level 2 (which is a lot of air) 4 noise passes makes it disappear. On level 1, 3 noise passes sound silent, whereas the low roll off nothing really changed.
Low Rolloff’s job is a little magic. It suppresses, as it says on the tin, thumps, earthquakes, and thunder. Stuff that’s barely audible. However if you try to apply regular voice processing after Low Rolloff, you find that your voice is louder and brighter because there is no competition from the rumble.
That’s why it’s the first step in Audiobook Mastering.
So you decided to go with the Dead Cat. Good call. We’ve been using those to get rid of wind noises for centuries.
