spauq.core.metrics

Functions

signal_to_spatial_distortion_ratio(reference, ...[, ...])	param reference: reference signal, shape=(n_channels, n_samples)
signal_to_residual_distortion_ratio(reference, ...[, ...])	param reference: reference signal, shape=(n_channels, n_samples)
spauq_eval(reference, estimate, fs, *[, ...])	param reference: reference signal, shape=(n_channels, n_samples)

Module Contents

signal_to_spatial_distortion_ratio(reference, estimate, fs, *, return_framewise=False)

Parameters:

• reference (numpy.ndarray) – reference signal, shape=(n_channels, n_samples)

• estimate (numpy.ndarray) – reference signal, shape=(n_channels, n_samples)

• fs (int) – sampling rate in Hz

• return_framewise (bool) – whether to return the framewise evaluation result

Returns:

Median signal to spatial distortion ratio over frames in dB. An array of shape (n_frames,) if return_framewise is true.

Return type:

float or np.ndarray

signal_to_residual_distortion_ratio(reference, estimate, fs, *, return_framewise=False)

Parameters:

• reference (numpy.ndarray) – reference signal, shape=(n_channels, n_samples)

• estimate (numpy.ndarray) – reference signal, shape=(n_channels, n_samples)

• fs (int) – sampling rate in Hz

• return_framewise (bool) – whether to return the framewise evaluation result

Returns:

Median signal to residual distortion ratio over frames in dB. An array of shape (n_frames,) if return_framewise is true.

Return type:

float or np.ndarray

spauq_eval(reference, estimate, fs, *, return_framewise=False, return_cost=False, return_shift=False, return_scale=False, verbose=True, **kwargs)

Parameters:

• reference (numpy.ndarray) – reference signal, shape=(n_channels, n_samples)

• estimate (numpy.ndarray) – reference signal, shape=(n_channels, n_samples)

• fs (int) – sampling rate in Hz

• return_framewise (bool) – whether to return the framewise evaluation result

• return_cost (bool) – whether to return the estimation error (cost function)

• return_shift (bool) – whether to return the estimated shift parameters

• return_scale (bool) – whether to return the esitmated scale parameters

• verbose (bool) – whether to use verbose mode

• kwargs – additional keyword arguments to be passed to the internal function.

Returns:

A dictionary with the following keys:

• SSR (float or np.ndarray)

  Median signal to spatial distortion ratio in dB. An array of shape (n_frames,) if return_framewise is true.

• SRR (float or np.ndarray)

  Median signal to residual distortion ratio in dB. An array of shape (n_frames,) if return_framewise is true.

• cost (float or np.ndarray)

  Median cost function of the estimates over frames. An array of shape (n_frames,) if return_framewise is true.

• shift (np.ndarray)

  An array of shape (n_channels, n_channels,) of the elementwise median of the shift matrices over frames. An array of shape (n_channels, n_channels, n_frames,) if return_framewise is true.

• scale (np.ndarray)

  An array of shape (n_channels, n_channels,) of the elementwise median of the scale matrices over frames. An array of shape (n_channels, n_channels, n_frames,) if return_framewise is true.

Return type:

dict