Automatic noise reduction when recording?

I’ve been playing around with recording some live performances for eventual podcasting, and I’ve had some unexpected results. The performances are spoken-word with standard lectern and clip-on mikes in a moderately noisy environment, and I’m recording directly off the PA amp’s record-out jacks. I’ve recorded some performances directly on my laptop with Audacity and others with a digital recording device. One I recorded with both (using a splitter off the amp, so they were getting the same signal). The curious thing is that the noise levels on the performances recorded directly with Audacity are significantly lower than those recorded with the digital recorder and imported into Audacity. By significantly, I mean that the ones recorded with Audacity have total silence where the ones recorded with the digital recorder have ambient noise to about 50% of the clip level. In addition, if I watch the waveform as I start the recording in Audacity, it seems at first to record the ambient noise and then after a few seconds, the waveform drops to zero, as if it had figured out that what it was hearing was the background noise and automatically removed it. The recording where I used both devices on the same signal is quite striking in the difference.

I’ve searched the web and this forum, and I can’t find anything about Audacity doing some sort of adaptive automatic noise removal during recording, but that’s what it seems like it’s doing. Does anybody know anything about this? Am I crazy?



<<<Am I crazy?>>>

I hope so. Sane people are booooooriiiing.


Tell us a lot about that laptop.

It’s not Audacity, by the way. Audacity doesn’t apply filters in real time…At all.


OK. Time to do some more experiments. It’s just a Dell D420 laptop. I’m sure I have an 1/8" to 1/8" patch cable around somewhere, so I’m going to try to put the output of the digital recorder into the laptop line in and see if I get the same effect. Maybe I’m hallucinating, but it’s pretty clear by playing the recordings I made from the same signal that one is significantly different than the other. I’m sure it’ll turn out to be a simple explanation, but right now I’m stumped. If you want to hear the difference, you can go to and listen to the two samples there. It seems like one is from an omni-directional mike in the center of the room and the other from a unidirectional mike right on the speaker, but I think they were getting the same signal out of the amp. I’ll try to repeat the setup again soon; probably for some reason the line-in on the digital recorder didn’t work and it was just using the built-in mike or something.

However, if anybody else has any ideas, I’d love to hear them.



Multiple Problems.

The first recording, the “live, bright one,” is overloading whatever is being used to record it. The recording has extremely high overall gain and volume boost and tries to reduce the volume on the fly when the performer talks. In broadcasting, this is called “pumping.” You can hear the background noise pump up and down with speech. You might get something like this if you have the Line-Out of a sound mixer plugged right into the Mic-In of a laptop. That’s a 1000 to one mismatch and will sound terrible.

The other one has a real-time level sense keyer running–or possibly some other damage. There are several problems with that. It’s possible for the performer to drop their volume a little and fall under the threshold of the recorder. That work will be lost. The other problem is background noise is not intended to go to zero. I’ve heard interviews try to do that and it sounds very weird. There are Hollywood movie sound departments whose job is to put background sound back in.

One of the production people in our building once tried recording in a completely echo-free room. They had to bring plywood sheets in because they couldn’t get a natural recording otherwise.

So you haven’t produced a good, natural recording yet.

I’m betting if I walked in with my Mac PowerBook and plugged it into the house mixer, I would get a perfect recording. If I had a PC Deskside machine with multiple sound inputs I could do it, too, or any computer with a good USB sound interface. I think you have multiple sound level mismatches and possibly actual electronic damage.


It’s not Audacity doing that, it’s your sound card.

I’ve noticed that quite a few laptops have “noise reduction” and/or Automatic Gain Control (AGC) built in, and switched on by default. While these features can be useful for VOIP (internet telephone), they make a complete mess of any attempts at recording. You need to find out how to switch them off. On the laptops that I have seen this, there has been an icon near the clock that opens a control panel for these audio “enhancements”.

Unfortunately, when you switch these “enhancements” off, you will probably be able to hear just how poor quality the audio input really is.

For recording with a laptop from a mixing desk, even the cheap USB line level interfaces (such as the Behringer UCA 202) are a massive improvement on the built in sound card.


…in a PC.


Yes, I hear that Mac laptops have quite good sound cards, but this is “just a Dell D420 laptop”.
Also, a good friend of mine has a Toshiba that’s about 5 or 6 years old now, and the on-board sound card is very nice.
(so much for technology always getting better).