One of the things I enjoy most on recordings is the time spent doing some design on the recordings setup and scouting the recording locations before removing the dust from the recorder. Often, the time allocated in this area pays-off in the amount of processing the recordings will need afterwards, letting you focus instead on making things sound good rather than fixing what was wrong in the first place and even possibly making it worse.
Acting over the recording space, looking for larger, wider and higher areas, using acoustics to your advantage and carefully choosing your microphones depending on the environment’s frequency response and pattern rejection are specific factors that have a huge impact on the kind of sound you’ll get out of the recordings.
Of course there might be aspects out of your scope of control, unexpected events you didn’t plan for etc… that have an impact and introduce unwanted noise in your recordings. Some of the most commonly known are related to traffic and other sources of low frequency rumble. In these cases, applying a combination of high pass filtering, multiband compression and noise reduction is usually a good option. Remember recording enough of unwanted noise so that you can use it as a pattern to remove it as much as possible from the source material in your noise reduction software of choice. Be gentle, avoid artifacts and over-processing.
Other ‘noise’ sources have to do with the resonating frequencies inherently exposed by standing waves in your recording space. This ‘sonic print’ might make your recordings sound out of context in most situations. In order to reduce the effect we need to get these frequencies, which are easily calculated by the next formula:
f1 = V / 2d, Where:
f1 = Fundamental frequency in Hz
V = 343 m/s
d = (length, width, and height in m)
The harmonics/modes are related to the fundamental frequency by the next formula: fn= n*f1. Where ‘n’ is the harmonic order. Frequencies would be calculated for all room dimensions.
Thus, if you are recording in a small space of 2x2x2 meters, result for one of the dimensions would be:
Fundamental Frequency -> f1 = (343/2*2) = 85.75 Hz
Second mode -> f2 = 85.75 Hz * 2 = 171.5 Hz
Third mode -> f3 = 85.75 Hz * 3 = 257.25 Hz
Fourth mode -> f4 = 85.75 Hz * 4 = 343 Hz
The frequencies above have a high chance to colour the signal you want to record, specially if this signal has some similar components in the spectrum. If possible, try to record in a larger space so the rooms fundamental frequency falls out the hearing range and its harmonics have a significantly lower amplitude. This will give you a cleaner recording, as first reflections and the sonic print of the space will be much less apparent. If that’s not an option, using multiband compression to control the low frequency content below the 3 or 4 first harmonics helps out most times. Also, running the signal through de-reverberator processors might help reduce reflections in the whole spectrum.
All in all, if you find out that noise is too much, you can always change plans and record a sound you didn’t expect having in the first place!
What other noise sources do you typically face and what methods do you use to manage noise in your recordings?