Thursday, April 25, 2013

So Noise...

I talked about it a little bit in the last post, but as my signal processing professor stressed, this is one of the most important definitions in signal processing, acoustics, vibrations, and image processing. That along with the reason sinusoids are studied (DOH, that's another blog entry) is the foundation of these sciences.

The definition I locked onto in the last post was that Noise was any unwanted sound/signal/vibration/data/image (plenty of unwanted images on the internet...). I also stated that this is a completely relative definition. In my lab, when I'm working on measuring some acoustic phenomenon, and my neighbor in the lab is listening to music through his ear phones so loud that I can hear it (with is horrific for his hearing) that music, to me, is noise, because it will be an unwanted artifact in my measurements. To him, it's part of how he focuses on his work, perhaps making him more productive... but also probably causing error in his measurements which makes it noise to him, as well!

It's also interesting that we redefine noise as the sound we might be looking for when trying to determine a problem with mechanical systems. For example: The noise my car is making when I hit the brakes. In the terms of that state of mind, when trying to diagnose that phenomenon, it now becomes the sound you're looking for rather than noise. In that case, if the engine is knocking, or an excessive amount of tire noise is present, those affects become noise, and the brake squeal becomes a sound phenomenon.

This means, by definition, noise is completely subjective based on the experimenter. Studies have been done for years trying to describe, in general, what noise is to the average person. The interesting thing is, that one of the things we typically call noise, namely white noise (defined as a random signal with a flat power spectrum), is actually perceived by most humans as a pleasant sound, the absence of which causes distress. This is why some people are uncomfortable when they leave the confines of a city and go out in the country. The ambient sound level, also called background noise, a very ironic term, is gone. These same people are very uncomfortable when in an anechoic room, because the only thing they have to listen to is all the stuff going on in their own bodies.

Humans, in general find blended sounds pleasing, and sharp sounds displeasing. Without getting too much into psychoacoustics (which is a real thing, go look on wikipedia), sudden sounds, impulsive noise, tonal noise, and poorly blended music all tend to have a negative affect on people. This is why sharp sudden tones (in the 1-4 kHz range, the most sensitive frequency range of the human ear), are used as warning sirens in buildings and on emergency vehicles.

However, to an experimenter, noise in your system is whatever you define it to be. You have to be very well versed in your system in order to diagnose this fact. If you're trying to work with an electric motor, and you've grounding issues, 60 Hz (or 50 Hz for those of you not in North America) is really bad. Not only is this going to be a frequency of interest, but it's also going to be a huge source of error in your measurements.

The last little bit of info I want to give tonight is this: Because noise is whatever the experimenter defines it to be, very few other people will be able to help you diagnose noise problems without you giving them a very well worded definition. The person who knows a set of data taken from a group of experiments is the person who took the data. This is why I prefer to take almost all of my own data, because then I can think backward and forward what all of the possible sources of noise might be. 

Tonight I'll leave you folks with a question... I spoke about sinusoids a couple times, why are they important? (Full circle... BWAHAHAHAHAHA!)