Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 13 views

    Slides STFTMoodle

    Uploaded by

    Rui Rodrigues
    The STFT provides three representations of a signal: 1. The original sound signal 2. A time signal representation 3. A time-frequency representation obtained by applying the Fourier Transform to short segments of the time signal. The STFT results in a 2D representation with frequency on the vertical axis and time on the horizontal axis, showing the evolution of frequency content over time. The color represents the amplitude of each frequency component, with blue being the highest.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Slides STFTMoodle

    Uploaded by

    Rui Rodrigues
    13 views7 pages
    The STFT provides three representations of a signal: 1. The original sound signal 2. A time signal representation 3. A time-frequency representation obtained by applying the Fourier Transform to short segments of the time signal. The STFT results in a 2D representation with frequency on the vertical axis and time on the horizontal axis, showing the evolution of frequency content over time. The color represents the amplitude of each frequency component, with blue being the highest.

    Original Description:

    short time fourier transform

    Original Title

    slidesSTFTMoodle

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The STFT provides three representations of a signal: 1. The original sound signal 2. A time signal representation 3. A time-frequency representation obtained by applying the Fourier Transform to short segments of the time signal. The STFT results in a 2D representation with frequency on the vertical axis and time on the horizontal axis, showing the evolution of frequency content over time. The color represents the amplitude of each frequency component, with blue being the highest.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    13 views7 pages

    Slides STFTMoodle

    Uploaded by

    Rui Rodrigues
    The STFT provides three representations of a signal: 1. The original sound signal 2. A time signal representation 3. A time-frequency representation obtained by applying the Fourier Transform to short segments of the time signal. The STFT results in a 2D representation with frequency on the vertical axis and time on the horizontal axis, showing the evolution of frequency content over time. The color represents the amplitude of each frequency component, with blue being the highest.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 7

    The Short Time Fourier Transform, STFT

    Three representations of the same signal:


    I Sound: Yes
    I Two graphical representations: time signal e time-frequency
    representation
    Fourier Series: basic idea
    Each periodic function, with period T, can be written as the ’sum’
    of sinusoids with periods dividing T.
    +∞   X+∞  
    X 2kπ 2kπ
    f (x) = a0 + ak cos x + bk sin x
    T T
    k=1 k=1
    Euler’s formula

    e iθ = cos θ + i sin θ
    e iθ + e −iθ
    cos θ =
    2
    e iθ − e −iθ
    sin θ =
    2i
    +∞   X +∞  
    X 2kπ 2kπ
    f (x) = a0 + ak cos x + bk sin x
    T T
    k=1 k=1
    +∞
    X 2kπ
    f (x) = Ck e i T
    x

    k=−∞

    Ck are complex numbers.


    The Fourier Transform of a discrete signal
    I f = (f (tk ))k=0,...,N−1 , time representation.
    tk = (k − 1)∆, where ∆ is a positive number.
    I Discrete Fourier Transform (DFT):

    DFT (f ) = (fˆk )k=0,...,N−1

    I The (fˆk )k=0,...,N−1 satisfy

    N−1
    X −2kiπ
    fˆk = tp e N

    p=0

    and are, in general, complex numbers.


    I The meaning of the (fˆk )k=0,...,N−1 is given by the Inverse
    Discrete Fourier Transform (IDFT)
    N−1
    1 X ˆ i 2kπ p
    f (tp ) = fk e N
    N
    k=0
    Some Fourier Transform examples
    STFT: Short Time Fourier Transform
    To get the STFT of a signal, we apply the Fourier Transform to
    short time signal segments. In general there can be an overlap
    between the signal segments.

    The result is a sequence of Fourier Transforms, one for each


    segment.
    In the discrete case, as each transfor is a vector, the STFT is a 2-D.
    In the case of a real signal, the number of frequencies is (FFT size
    +1)/2 +1.. The number of segments where we apply the DFT
    (the STFT time dimension) is approximately the L/(FFT size -
    Overlap), where L is the length of the time-signal. Therefore, there
    is an important reduction in the time dimension whene we apply
    the STFT.
    The color represent
    the STFT absolut value of each frequency component: blue is the
    hights value, followed by the green and then yellow.
    In the image above, the vertical axis represents the frequencies and
    the horizontal axis represents time.
    In the STFT image the horizontal dimension was amplified to
    coincide with the corresponding dimension of the original signal.

    You might also like