class FilterBank(fs: float, channels: int, f_min: float, f_max: float, order: int, filter_type: Literal['butterworth', 'gammatone', 'sos'] = 'butterworth', **kwargs)[source]

Bases: ABC

A filter bank for decomposing signals into frequency components.

This class decomposes input signals into frequency components using filter banks. Supported filter types are Butterworth, Gammatone, and second-order section (SOS) filters. The filter bank automatically computes center frequencies and frequency bands for each channel.

Code Example:

import numpy as np
from spikify.filters import FilterBank
fs = 1000
signal = np.random.randn(1000)
filterbank = FilterBank(fs=fs, channels=4, f_min=100, f_max=800, order=2)
freq_components = filterbank.decompose(signal)
Parameters:

  • fs (float) – Sampling frequency of the input signal.

  • channels (int) – Number of filter channels.

  • f_min (float) – Minimum frequency for the filter bank.

  • f_max (float) – Maximum frequency for the filter bank.

  • order (int) – Order of the filters.

  • filter_type (str) – Type of filter (‘butterworth’, ‘gammatone’, ‘sos’).

  • kwargs (dict) – Additional filter parameters.

Raises:

ValueError – If filter_type is not supported.

Attributes:
center_frequencies

Center frequencies for each filter channel.

Methods

decompose(signal)

Decompose input signal into frequency components using the filter bank.
decompose(signal: ndarray) → ndarray[source]

Decompose input signal into frequency components using the filter bank.

This method applies each filter in the bank to the input signal and returns the filtered outputs for all channels.

Parameters:

signal (numpy.ndarray) – Input signal to be decomposed. Should be a 1D or 2D numpy array. If 2D, shape should be (timestamps, features).

Returns:

Array of filtered signals with shape (timestamps, channels, features).

Return type:

numpy.ndarray

property center_frequencies: ndarray

Center frequencies for each filter channel.

This property returns the center frequency for each channel in the filter bank.

Returns:

Array of center frequencies.

Return type:

numpy.ndarray