DFT

Class for performing the Discrete Fourier Transform (DFT) and inverse DFT for real signals, including multichannel. It is also possible to specific an analysis or synthesis window.

When available, it is possible to use the pyfftw or mkl_fft packages. Otherwise the default is to use numpy.fft.rfft/numpy.fft.irfft.

More on pyfftw can be read here: https://pyfftw.readthedocs.io/en/latest/index.html

More on mkl_fft can be read here: https://github.com/IntelPython/mkl_fft

class pyroomacoustics.transform.dft.DFT(nfft, D=1, analysis_window=None, synthesis_window=None, transform='numpy', axis=0, precision='double', bits=None)

Bases: object

Class for performing the Discrete Fourier Transform (DFT) of real signals.

X

Real DFT computed by analysis.

Type

numpy array

x

IDFT computed by synthesis.

Type

numpy array

nfft

FFT size.

Type

int

D

Number of channels.

Type

int

transform

Which FFT package will be used.

Type

str

analysis_window

Window to be applied before DFT.

Type

numpy array

synthesis_window

Window to be applied after inverse DFT.

Type

numpy array

axis

Axis over which to compute the FFT.

Type

int

precision, bits

How many precision bits to use for the input. Twice the amount will be used for complex spectrum.

Type

string, np.float32, np.float64, np.complex64, np.complex128

Parameters

  • nfft (int) – FFT size.

  • D (int, optional) – Number of channels. Default is 1.

  • analysis_window (numpy array, optional) – Window to be applied before DFT. Default is no window.

  • synthesis_window (numpy array, optional) – Window to be applied after inverse DFT. Default is no window.

  • transform (str, optional) – which FFT package to use: numpy, pyfftw, or mkl. Default is numpy.

  • axis (int, optional) – Axis over which to compute the FFT. Default is first axis.

  • precision (string, np.float32, np.float64, np.complex64, np.complex128, optional) – How many precision bits to use for the input. If ‘single’/np.float32/np.complex64, 32 bits for real inputs or 64 for complex spectrum. Otherwise, cast to 64 bits for real inputs or 128 for complex spectrum (default).

analysis(x)

Perform frequency analysis of a real input using DFT.

Parameters

x (numpy array) – Real signal in time domain.

Returns

DFT of input.

Return type

numpy array

synthesis(X=None)

Perform time synthesis of frequency domain to real signal using the inverse DFT.

Parameters

X (numpy array, optional) – Complex signal in frequency domain. Default is to use DFT computed from analysis.

Returns

IDFT of self.X or input if given.

Return type

numpy array