Noise is unwanted information.It is parasite data that interferes with relevant data and therefore it is aimed at keeping it as low as possible.
There are a lot of noise shapes like video, audio, electric interference, radio interference, etc.
We will be peeking a bit more into audio noise. This may be roughly divided into mechanical, electrical and media/storage noise.
- In a recording process, especially when recording non-electronic instruments, we may deal with sounds that are not directly related to the instrument (or vocals). This may be ambient noise, audience (if recording live), moving chairs, moving performers, etc.
- Electrical noise is the one that is generated by electronic devices. No electronic component is ideal and has some unwanted characteristics. Assuming that we have tens or hundreds of these components in a circuit, that just adds to the trouble. To reduce the noise and boost the overall performance of an audio device, it is therefore vital to use as good quality components as it is rationally allowed.
Another type of electrical noise is hum. This is when we hear 50 or 60 Hz (depends on our location and therefore mains power standard) in the device. This appears in the devices plugged into the mains power network.
- There are different storage media available, either analog (magnetic tape, vinyl record) or digital (CD, DAT, digital files). While on the analog media we have to deal with tape noise and cracking and popping of the record, on the digital media we are limited with the quality of AD/DA converters, their setting and type of file saving.
Signal-to-noise ratio (SNR) is information that tells us how much noise a device produces compared to the relevant signal and it is measured/calculated in dB.
This is what noise looks like compared to the relevant signal (an overdone illustration). On the left there is a bad SNR – the noise is pretty large to the signal itself, while on the right this ratio is much better.
SNR is therefore expressed in dB: the higher the value, the better it is. To have some reference of what numbers we are dealing with, at 80 dB SNR we can bearly hear any noise. Below that we begin to perceive the noise as a part of the sound. Whatever goes above 80 dB is just fine. Over 90 dB there is a little chance that you will be able to perceive anything else than the relevant signal itself.
SNR is not vital data when observing characteristics of an audio system. Obviously it is better to keep it as high as possible, but it doesn't tell anything more about the performance of the system than how much noise it produces.