01-16-2014 08:41 PM
Hi,
I am trying to figure out how to allow a user to add white noise to a signal, the idea is that the user would enter the noise ratio in db which I would then apply to a signal measurement before displaying it. I looked at the add gaussian white noise vis in LV but I am not sure how to go from the noise ratio that the user enters to the seed and standard deviation inputs for the function.
I tried to look online but all I found was ways to add noise based on amplitude, is there a way to convert from the db ratio to amplitude given that I do not know the amplitude information of my original signal before hand since it is being sampled in real-time. Any information about this would be greatly appriciated i realize that I might be mis-understanding the way signal/noise processing really works.
Thank you,
A
01-16-2014 08:54 PM
Well, you certainly need to know the signal amplitude if you want to create a specific signal to noise ratio.
Are you going to add a random value to each point of signal or will you have a block of signals at a time? You could set the amplitude of the noise based on the average of some previous values or base it on the full scale range of the receiver.
Lynn
01-16-2014 09:48 PM
dB is a dimensionless relative measure, you need a reference intensity in order to calculate it
Often times in electronics you see things referenced to 1 mW, called dBm = 10 log10(P / 1mW), so P = 10^(dBm/20)
http://en.wikipedia.org/wiki/DBm
For sound dB SPL = 20 log10(I / I0), I (intensity) is usually expressed Pascalls I0 is nearly always referencec to human hearing threshold, which is 2 * 10^-5 Pa. Multiply by your mic gain to convert to from pressure to volts.
01-17-2014 07:50 AM
For the first implementation, I will have the full signal provided to me which I will then allow the user to adjust the level of noise they want to add to it.
So form this would i be correct if I try to get the average amplitude of the signal full signal and use that for the ratio?
01-17-2014 08:11 AM
You or your user need to decide if that is "correct." It seems like a reasonable starting point to me.
Lynn