Islamic university of Gaza Faculty of engineering Electrical engineering dept. Submitted to: Dr.Hatem Alaidy Submitted by: Ola Hajjaj 2003- 3005 Tahleel Abu seedo 2003-4240 Short T ime F ourier T ransform
Mar 28, 2015
Islamic university of Gaza
Faculty of engineering
Electrical engineering dept.
Submitted to:
Dr.Hatem Alaidy
Submitted by:
Ola Hajjaj 2003-3005
Tahleel Abu seedo 2003-4240
ShortT ime F ourier T ransform
Contents
History
The Fourier Transform
Why STFT
Formula of STFT
Windows definition
STFT windows
Resolution concept
Comparisons
Inverse of STFT
Application for STFT
Conclusion
History of19th century, J. Fourier, reach to the formula of periodic
function as an infinite sum of periodic complex exponential functions.
Many years after, non-periodic functions were generalized.
Then periodic & non-periodic discrete time signals were known.
In 1965, (FFT) was known.
The Fourier Transform
DFT: used When fs>=2fm, and the transformed signal is symmetrical.
FT: decomposes a signal to complex exponential functions of different frequencies
FFT: to reduce the no. of multiplications in DFT.
STFT
X(f)=-∞∫ ∞ x(t).e-2j∏ft dt……..(1)
x(t)= -∞∫ ∞ X(f). e-2j∏ft df…...(2)
WhyIt gives a suitable description for the local change in frequency content because the frequency component which defined by FT have infinite time support.
STFT provides a means of joint time-frequency analysis.
Continue.
In STFT, the signal is divided into small enough segments.
For this purpose, a window function "w" is chosen. The width of this window must be equal to the segment of the signal.
Formula of
x(t) is the signal itself,
w(t) is the window function, and
* is the complex conjugate
The STFT of the signal is the FT of the signal multiplied by a window function.
STFTx(w)(,f)=t∫[x(t).w*(t- ).e-2j∏ft dt……………(3)
Note That:
The STFT of a signal x (n) is a function of two variables: time and frequency.
Windows
-real and symmetric .
-Function with zero-valued outside of some chosen interval .
Definition
Windows Properties
Trade-off of time versus frequency resolution.
Detectability of sinusoidal components.
Zero phase window.
Hanning window Gaussian windows
W(t)
Windows of
Transforming steps in
This window function is located at the beginning of the signal At (t=0).
The window function will overlap with the first T/2 seconds of the original signal
The window function and the signal are then multiplied.
Taking the FT of the product.
The window would be shifted by t1 to a new location multiplying with the signal.
Repeat from step 3 Until the end of the signal.
Window & Resolution
STFT has a fixed resolution.
The width of the windowing function relates to the how the signal is represented.
It determines whether there is good frequency resolution or good time resolution
Narrow window
Narrowband and Wideband Transforms.
good time resolution, poor frequency resolution.
Wide window
good frequency resolution, poor time resolution.
Spectrogram
Resolution Explanation
The Gaussian window function in the form:
w(t)=exp(-a*(t^2)/2);
Range of freq. Separated peaks in
time
Case 1:
Case 2:
Much better resolution Not se
parat
ed p
eaks
Case 3:
High frequency resolution Low
tim
e re
solu
tion
Inverse of
Time-Frequency Trade-off
Comparisons
The signal multiplied by a window function.
Transform is a function of both time and frequency
There is resolution problem in the frequency domain
Window is of finite length
Its window is exp{jwt} function, from minus infinity to plus infinity
no resolution problems in freq. domain
One domain only
One window
Application for
The problem of
No exact time-frequency representation of a signal
Resolution problem, time intervals in which certain band of frequencies exist.
Wavelet transform (or multi resolution analysis) high-frequency gives good time resolution for events, and good frequency resolution for low-frequency events, which is the type of analysis best suited for many real signals.
The Solution:
Conclusion
STFT is a Fourier related transform & it is a Function of two variable (time & frequency).
Used to determined the freq. and phase content of local section of a signal over time.
It deals with two windows (hanning & Gaussian).
There is a relation between window and resolution .
Thank you for listening.