How to avoid noise in a voice recording

Question

I'm recording for a podcast with a Rodhe NT2 microphone and I'm getting a lot of noise in the background, although the room is pretty quiet. It looks like this mic doesn't pick up my voice very loud and I have to increase the gain to the max on the audio interface (a Beringher Uphoria UMC22).

So, I tried to record with a Shure SM58, which is more directional. This time I can keep the gain in the middle and get a good volume, but I still get noise, although less than before.

I would like to get rid of the background noise completely, as I like to work on my recordings and cut or removing parts, move other parts around and so on. So, I don't want the user to notice all these cuts and pastes, by noticing all the interruptions in the background noise.

I also tried to reduce the background noise with Logic Pro and I used plugins like Expander, Denoiser, Noise Gate or Speech Enancher... but not with great results so far.

Online I see a lot of videos with perfectly clear audio. It seems like for other people it's easy to get rid of the noise, and I really don't get what I'm doing wrong. Is it the wrong mic that I'm using? what else can I try?

Answer 1

what else can I try?

Use a multiband dynamics processor that can be set up as both a compressor and expander, and use it as a multiband gate. For each frequency band, particularly high frequencies, attenuate everything that's below a threshold level. I don't know if Logic has such a plugin, but Ableton Live has. I use Ableton's multiband dynamics plugin, and it gets rid of practically all background noise. There's even a ready-made preset called "Reduce Ambience" for this. If Logic doesn't come with such a thing, I'm sure there are third-party plugins you can use.

Ok, I see that this annoyed some people, because it suggests an easy fix with a magical trick instead of doing things "properly". And of course, learn your basic recording skills like setting up mics, move to a quieter area, build a recording room etc. but if you can't or aren't interested in improving those aspects any more for some reason, a multiband gate can help.

Answer 2

I had a similar problem, and realized that my connection to the recording device (my computer) was bad. I was using an amplifier/mixer connected directly to the PC's built-in audio interface (i.e. not using USB), and couldn't figure out where the noise was coming from. By using a real "pro-sumer" audio interface (with its own built-in phantom power & amp), a lot of noise was removed — in my case, the noise wasn't in the room, but in the circuitry of my equipment!

The thing that clued me in originally was that using an H4N together with the same microphone in the same environment produced much less noisy recordings. I first tried replacing the audio interface in my PC with a PCI card. When that didn't work, I got the fully external audio interface linked above.

I really don't know whether the noise was in my PC's power source, audio circuits, or the cheap mixer I was using. This change got rid of 50% of my noise, but I'm not an audio expert (yet), so do your own research before you spend your own money. :-)

PS: The H4N had great audio quality and on-the-go convenience, but its software is a pain to use! It felt like using one of those old mp3 players. The only reason I replaced it with the Scarlett is that I lost it; but I like the Scarlett better.

Answer 3

Any decent DAW will have a Noise Reduction option. They usually work by sampling the noise floor from a few seconds of audio just recording the background noise, then using that profile to process the noise out.

Of course that should be done after all the other efforts suggested elsewhere to reduce the noise floor in the first place, get your mic proximity right, and set your mic up optimally.

Answer 4

This answer's primary motif is the difference between Random and Systematic Errors in a Scientific Experiment. Random errors are pretty random as the name says and cannot be correct once the data collection phase is over, while Systematic errors are those that are almost constant and can be corrected later after the experiment's data collection phase is over. For example, A random error maybe something like the person conducting the experiment observed the apparatus from an angle that cause the Parallax to give him wrong reading, while a Systematic error maybe a zero error in the measuring apparatus and can be subtracted from all the reading later to correct it.

in short: Bring down external noise to an almost constant-ish one so that you can eliminate it with audio processing, analogous to Random & Systematic Errors in Scienctific experiments.

Try removing as much removable noise as possible. Ask people around you to not speak until you finish recording, turn off fans, and Air Conditioner, etc..

Try mechanical methods like a pop filter, Shock mount. So you've mentioned sound pickup patterns of mics so you've got that covered, Know the best spot to place your Mic.

So finally you are only left with noise that is not removable physically and is pretty constant in frequency distribution and Levels, at this point you can apply both a dynamic noise gate or an expander or a denoiser and get rid of the noise.

Hope this helps you a little if not fully.Because this is pretty much what I do when my friends send me their recordings to put together with other. They record with earphones and Speakers and what not! But all that I tell them is to ask their family members to stay quite till they finish recording, ask them to make sure fans are turned off (and their potato pcs are which make a lot of noise) and the recordings they send have a pretty constant noise level. Most of the time expansion + dynamic noise gate suffices. I've never used the noise-floor reduction method so far even once to denoise their recordings, its a good method to learn but maybe not as necessary if it seems involved.

But if noise is not the primary goal but just that the audience don't notice the cuts, then I think ideally all you need to do is run faders where the cuts are happening.