DSP Blockset

Real Cepstrum

Compute the real cepstrum of an input.

Library

Transforms

Description

The Real Cepstrum block computes the real cepstrum of each channel in the M-by-N input matrix, u. For both sample-based and frame-based inputs, the block assumes that each input column is a frame containing M consecutive samples from an independent channel.

The output is a real M_o-by-N matrix, where M_o is specified by the FFT length parameter. Each output column contains the length-M_o real cepstrum of the corresponding input column.

y = real(ifft(log(abs(fft(u,Mo)))))    % Equivalent MATLAB code

or, more compactly,

y = rceps(u,Mo)

When the Inherit FFT length from input port dimensions check box is selected, the output frame size matches the input frame size (M_o=M). In this case, a sample-based length-M row vector input is processed as a single channel (i.e., as an M-by-1 column vector), and the output is a length-M row vector. A 1-D vector input is always processed as a single channel, and the output is a 1-D vector.

The output is always sample-based, and the output port rate is the same as the input port rate.

Dialog Box

Inherit FFT length from input port dimensions

When selected, matches the output frame size to the input frame size.

FFT length

The number of frequency points at which to compute the FFT, which is also the output frame size, M_o. This parameter is available when Inherit FFT length from input port dimensions is not selected.

See Also

Complex Cepstrum	DSP Blockset
DCT	DSP Blockset
FFT	DSP Blockset
rceps	Signal Processing Toolbox

Random Source

Reciprocal Condition