Minimum Phase Filter Design

Above we used the Hilbert transform to find the imaginary part of an analytic signal from its real part. A closely related application of the Hilbert transform is contructing a minimum phase frequency response from an amplitude response.

Let $ H(e^{j\omega})$ denote a desired complex, minimum-phase frequency response,

$\displaystyle H(e^{j\omega}) \isdef G(\omega)e^{j\Theta(\omega)},

and suppose we have only the amplitude response

$\displaystyle G(\omega) \isdef \left\vert H(e^{j\omega})\right\vert.

Then the phase response $ \Theta(\omega)$ can be computed as the Hilbert transform of $ \ln G(\omega)$. This can be seen by inspecting the log frequency response:

$\displaystyle \ln H(e^{j\omega}) = \ln G(\omega) + j\Theta(\omega)

If $ \Theta$ is computed from $ G$ by the Hilbert transform, then $ \ln
H(e^{j\omega})$ is an ``analytic signal'' in the frequency domain. Therefore, it has no ``negative times''. The time domain signal corresponding to a log spectrum is called the cepstrum. It is shown in §E.9 below that a frequency response is minimum phase if and only if the corresponding cepstrum is causal [187, Ch. 10], [247, Ch. 11].

Next Section:
Minimum Phase and Causal Cepstra
Previous Section:
Generalized Window Method