Filter Types

pyfar.dsp.filter contains different filter types that are briefly introduced in the following. The filters can be used to directly filter a Signal or can return a Filter object described in Filter classes.

High-pass, low-pass, band-pass, and band-stop filters

These are the classic filters that are wrapped from scipy.signal and available from the functions butterworth, bessel, chebyshev1, chebyshev2, and elliptic.

Linkwitz-Riley cross-over

The function crossover returns Linkwitz-Riley cross-over filters that are often used in loudspeaker design.

Filter banks

Filter banks are commonly used in audio and acoustics signal processing, pyfar contains the following types of filter banks:

• The fractional_octave_bands are often used for calculating room acoustic parameters

• The reconstructing_fractional_octave_bands can be used if a perfect reconstruction is required, e.g., in room acoustical simulations.

• The auditory gammatone filters can be used for binaural modeling.

The corresponding center frequencies are accessible via fractional_octave_frequencies and erb_frequencies.

Parametric equalizer

The high_shelve, low_shelve, and bell filters shown on the left are specific filters for digital audio signal processing and are often used for audio effects and loudspeaker or room compensation. Bell filters manipulate the magnitude response around a center-frequency. Low- and high-shelve filters manipulate the magnitude response below and above a characteristic frequency. The cascaded shelving filters low_shelve_cascade and high_shelve_cascade shown on the right can be used to generate filters with a user definable slope given in dB per octaves within a certain frequency region.