Cepstrum of a list:

Cepstrum of an Audio object:

Plot of the power cepstrum:

Real cepstrum of a list:

Cepstrum of a 2D list:

Compute the cepstrum of a Sound :

Cepstrum of a multichannel Audio object:

The cepstrum is computed separately on each channel:

Compute the cepstrum of the audio track of a video object:

Cepstrum of an Image object:

Plot of the power cepstrum:

Cepstrum of a multichannel image:

The cepstrum is computed separately on each channel:

Detect the effect of a comb filer on a signal:

The signal only has two sinusoidal components:

A comb filter with delay of 31 samples is applied:

It is not easy to identify the periodicity of the comb filter by using conventional spectral analysis:

Since the cepstrum of a convolution is the sum of the cepstra of the two components, it is easier to identify the peak caused by the comb filter:

Measure the time constant of an echo:

Compute the logarithm of the cepstrum, and discard the second half (the real cepstrum is symmetric):

Find the peaks:

Plot the result:

Select the position of the biggest peak:

Compute the period by dividing the quefrency by the sample rate:

Detect the pitch of a recording:

In harmonic sounds, the pitch does not correspond to the biggest peak in the spectrum:

Compute the cepstrum and discard the symmetric part:

Find the peaks and display the result:

The peak corresponding to zero quefrency is discarded, and the biggest peak is selected:

Compute the fundamental frequency by dividing the sample rate by the quefrency:

Check the result: