Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

signal normalization

Status
Not open for further replies.

Biruntha

Junior Member level 1
Joined
May 26, 2015
Messages
16
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Location
Jaffna, Srilanka
Activity points
102
Hi,
can anyone explain briefly what is signal normalization? why we need to normalization a signal?
 

Hi,

A bit more information please.

Are you talking about digital data, analog data like sensor or audio?

Klaus
 

I found this code to normalize. please explain how it works as steps.


Code dot - [expand]
1
2
3
4
5
if std(signal) < 1.0e-7
    signal= signal- mean(signal);   % neutralization
else
    signal= signal- mean(signal);   % neutralization
    signal= signal/ std(signal);    % normalization



Thankyou
 
Last edited by a moderator:

Audio peaks may be compressed or normalized with linear gain by post processing or pre-processing to meet desired standard output levels such as xVpp or x dBmV or x Vrms determined by the interface input. This may be required to limit modulation for example or normalize many signals to be mixed or maximize SNR.

this code does none of the above and uses 2 unknown constants, mean and std. But may elsewhere be trying to compute mean and std deviation.. It appears to use assume unsigned or rectified signals

To normalize implies measuring unsigned peaks then scaling by max value to get a gain to factorize every value (post-processing, linear) for streaming , adjusting gain by 1 dB increments with a quasi-peak detection method is used by Sony professional equipment.
 

Hi,

Your code makes that - during quiet phases - the DC offset is cancelled
And during loud phases additionally the volume is adjusted.

Klaus
 

Normalizing a signal x(n) = x(n) / (|x(n)|^2)

where the signal x(n) is a vector.

We do normalization to scale the signal magnitude to unity without changing the direction of the vector. This is done usually when the signal x(n) is too high or too low(close to zero).
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top