aspcol.soundfieldestimation.krr_stationary_mics_direction_regularized_changedip

aspcol.soundfieldestimation.krr_stationary_mics_direction_regularized_changedip(ir_mic, pos_mic, pos_eval, samplerate, c, reg_param, direction, beta)

Estimates the impulse responses at the evaluation points using kernel ridge regression.

Adds a directional weighting to regularize the sound field. This is done by just changing out the inner product in the RKHS to a weighted inner product, the same as was done in [koyamaSpatial2021]. The result is that the solution has the same form as the unregularized solution, but with a different kernel function.

Parameters:

  • ir_mic (np.ndarray of shape (num_mics, ir_len)) – The impulse responses measure at the microphones.

  • pos_mic (np.ndarray of shape (num_mics, 3)) – The position of the microphones.

  • pos_eval (np.ndarray of shape (num_eval, 3)) – The position of the evaluation points.

  • c (float) – The speed of sound.

  • reg_param (float) – The regularization parameter. Scales the regularization term in the optimization.

  • direction (np.ndarray of shape (1, 3)) – a unit vector describing the direction of the weighting. The direction should be from (0,0,0) towards the source.

  • beta (float) – The strength of the weighting. A larger value will give more regularization.

Returns:

ir_eval – The estimated impulse responses at the evaluation points.

Return type:

np.ndarray of shape (num_eval, ir_len)