def test_complex_impl_consistency():
if not is_torch_1_9_plus:
return
mat_th = torch.complex(torch.from_numpy(mat_np.real),
torch.from_numpy(mat_np.imag))
mat_ct = ComplexTensor(torch.from_numpy(mat_np.real),
torch.from_numpy(mat_np.imag))
bs = mat_th.shape[0]
rank = mat_th.shape[-1]
vec_th = torch.complex(torch.rand(bs, rank),
torch.rand(bs, rank)).type_as(mat_th)
vec_ct = ComplexTensor(vec_th.real, vec_th.imag)
for result_th, result_ct in (
(abs(mat_th), abs(mat_ct)),
(inverse(mat_th), inverse(mat_ct)),
(matmul(mat_th,
vec_th.unsqueeze(-1)), matmul(mat_ct, vec_ct.unsqueeze(-1))),
(solve(vec_th.unsqueeze(-1),
mat_th), solve(vec_ct.unsqueeze(-1), mat_ct)),
(
einsum("bec,bc->be", mat_th, vec_th),
einsum("bec,bc->be", mat_ct, vec_ct),
),
):
np.testing.assert_allclose(result_th.numpy(),
result_ct.numpy(),
atol=1e-6)
def get_rtf(
psd_speech,
psd_noise,
mode="power",
reference_vector: Union[int, torch.Tensor] = 0,
iterations: int = 3,
use_torch_solver: bool = True,
):
"""Calculate the relative transfer function (RTF)
Algorithm of power method:
1) rtf = reference_vector
2) for i in range(iterations):
rtf = (psd_noise^-1 @ psd_speech) @ rtf
rtf = rtf / ||rtf||_2 # this normalization can be skipped
3) rtf = psd_noise @ rtf
4) rtf = rtf / rtf[..., ref_channel, :]
Note: 4) Normalization at the reference channel is not performed here.
Args:
psd_speech (torch.complex64/ComplexTensor):
speech covariance matrix (..., F, C, C)
psd_noise (torch.complex64/ComplexTensor):
noise covariance matrix (..., F, C, C)
mode (str): one of ("power", "evd")
"power": power method
"evd": eigenvalue decomposition
reference_vector (torch.Tensor or int): (..., C) or scalar
iterations (int): number of iterations in power method
use_torch_solver (bool): Whether to use `solve` instead of `inverse`
Returns:
rtf (torch.complex64/ComplexTensor): (..., F, C, 1)
"""
if mode == "power":
if use_torch_solver:
phi = solve(psd_speech, psd_noise)
else:
phi = matmul(inverse(psd_noise), psd_speech)
rtf = (
phi[..., reference_vector, None]
if isinstance(reference_vector, int)
else matmul(phi, reference_vector[..., None, :, None])
)
for _ in range(iterations - 2):
rtf = matmul(phi, rtf)
# rtf = rtf / complex_norm(rtf, dim=-1, keepdim=True)
rtf = matmul(psd_speech, rtf)
elif mode == "evd":
assert (
is_torch_1_9_plus
and is_torch_complex_tensor(psd_speech)
and is_torch_complex_tensor(psd_noise)
)
e_vec = generalized_eigenvalue_decomposition(psd_speech, psd_noise)[1]
rtf = matmul(psd_noise, e_vec[..., -1, None])
else:
raise ValueError("Unknown mode: %s" % mode)
return rtf
def get_lcmv_vector_with_rtf(
psd_n: Union[torch.Tensor, ComplexTensor],
rtf_mat: Union[torch.Tensor, ComplexTensor],
reference_vector: Union[int, torch.Tensor, None] = None,
use_torch_solver: bool = True,
diagonal_loading: bool = True,
diag_eps: float = 1e-7,
eps: float = 1e-8,
) -> Union[torch.Tensor, ComplexTensor]:
"""Return the LCMV (Linearly Constrained Minimum Variance) vector
calculated with RTF:
h = (Npsd^-1 @ rtf_mat) @ (rtf_mat^H @ Npsd^-1 @ rtf_mat)^-1 @ p
Reference:
H. L. Van Trees, “Optimum array processing: Part IV of detection, estimation,
and modulation theory,” John Wiley & Sons, 2004. (Chapter 6.7)
Args:
psd_n (torch.complex64/ComplexTensor):
observation/noise covariance matrix (..., F, C, C)
rtf_mat (torch.complex64/ComplexTensor):
RTF matrix (..., F, C, num_spk)
reference_vector (torch.Tensor or int): (..., num_spk) or scalar
use_torch_solver (bool): Whether to use `solve` instead of `inverse`
diagonal_loading (bool): Whether to add a tiny term to the diagonal of psd_n
diag_eps (float):
eps (float):
Returns:
beamform_vector (torch.complex64/ComplexTensor): (..., F, C)
""" # noqa: H405, D205, D400
if diagonal_loading:
psd_n = tik_reg(psd_n, reg=diag_eps, eps=eps)
# numerator: (..., C_1, C_2) x (..., C_2, num_spk) -> (..., C_1, num_spk)
if use_torch_solver:
numerator = solve(rtf_mat, psd_n)
else:
numerator = matmul(inverse(psd_n), rtf_mat)
denominator = matmul(rtf_mat.conj().transpose(-1, -2), numerator)
if isinstance(reference_vector, int):
ws = inverse(denominator)[..., reference_vector, None]
else:
ws = solve(reference_vector, denominator)
beamforming_vector = matmul(numerator, ws).squeeze(-1)
return beamforming_vector
def get_WPD_filter(
Phi: Union[torch.Tensor, ComplexTensor],
Rf: Union[torch.Tensor, ComplexTensor],
reference_vector: torch.Tensor,
use_torch_solver: bool = True,
diagonal_loading: bool = True,
diag_eps: float = 1e-7,
eps: float = 1e-8,
) -> Union[torch.Tensor, ComplexTensor]:
"""Return the WPD vector.
WPD is the Weighted Power minimization Distortionless response
convolutional beamformer. As follows:
h = (Rf^-1 @ Phi_{xx}) / tr[(Rf^-1) @ Phi_{xx}] @ u
Reference:
T. Nakatani and K. Kinoshita, "A Unified Convolutional Beamformer
for Simultaneous Denoising and Dereverberation," in IEEE Signal
Processing Letters, vol. 26, no. 6, pp. 903-907, June 2019, doi:
10.1109/LSP.2019.2911179.
https://ieeexplore.ieee.org/document/8691481
Args:
Phi (torch.complex64/ComplexTensor): (B, F, (btaps+1) * C, (btaps+1) * C)
is the PSD of zero-padded speech [x^T(t,f) 0 ... 0]^T.
Rf (torch.complex64/ComplexTensor): (B, F, (btaps+1) * C, (btaps+1) * C)
is the power normalized spatio-temporal covariance matrix.
reference_vector (torch.Tensor): (B, (btaps+1) * C)
is the reference_vector.
use_torch_solver (bool): Whether to use `solve` instead of `inverse`
diagonal_loading (bool): Whether to add a tiny term to the diagonal of psd_n
diag_eps (float):
eps (float):
Returns:
filter_matrix (torch.complex64/ComplexTensor): (B, F, (btaps + 1) * C)
"""
if diagonal_loading:
Rf = tik_reg(Rf, reg=diag_eps, eps=eps)
# numerator: (..., C_1, C_2) x (..., C_2, C_3) -> (..., C_1, C_3)
if use_torch_solver:
numerator = solve(Phi, Rf)
else:
numerator = matmul(inverse(Rf), Phi)
# NOTE (wangyou): until PyTorch 1.9.0, torch.trace does not
# support bacth processing. Use FC.trace() as fallback.
# ws: (..., C, C) / (...,) -> (..., C, C)
ws = numerator / (FC.trace(numerator)[..., None, None] + eps)
# h: (..., F, C_1, C_2) x (..., C_2) -> (..., F, C_1)
beamform_vector = einsum("...fec,...c->...fe", ws, reference_vector)
# (B, F, (btaps + 1) * C)
return beamform_vector
def get_rtf(
psd_speech,
psd_noise,
reference_vector: Union[int, torch.Tensor, None] = None,
iterations: int = 3,
use_torch_solver: bool = True,
):
"""Calculate the relative transfer function (RTF) using the power method.
Algorithm:
1) rtf = reference_vector
2) for i in range(iterations):
rtf = (psd_noise^-1 @ psd_speech) @ rtf
rtf = rtf / ||rtf||_2 # this normalization can be skipped
3) rtf = psd_noise @ rtf
4) rtf = rtf / rtf[..., ref_channel, :]
Note: 4) Normalization at the reference channel is not performed here.
Args:
psd_speech (torch.complex64/ComplexTensor):
speech covariance matrix (..., F, C, C)
psd_noise (torch.complex64/ComplexTensor):
noise covariance matrix (..., F, C, C)
reference_vector (torch.Tensor or int): (..., C) or scalar
iterations (int): number of iterations in power method
use_torch_solver (bool): Whether to use `solve` instead of `inverse`
Returns:
rtf (torch.complex64/ComplexTensor): (..., F, C, 1)
"""
if use_torch_solver:
phi = solve(psd_speech, psd_noise)
else:
phi = matmul(inverse(psd_noise), psd_speech)
rtf = (phi[..., reference_vector,
None] if isinstance(reference_vector, int) else matmul(
phi, reference_vector[..., None, :, None]))
for _ in range(iterations - 2):
rtf = matmul(phi, rtf)
# rtf = rtf / complex_norm(rtf)
rtf = matmul(psd_speech, rtf)
return rtf
def get_mvdr_vector(
psd_s,
psd_n,
reference_vector: torch.Tensor,
use_torch_solver: bool = True,
diagonal_loading: bool = True,
diag_eps: float = 1e-7,
eps: float = 1e-8,
):
"""Return the MVDR (Minimum Variance Distortionless Response) vector:
h = (Npsd^-1 @ Spsd) / (Tr(Npsd^-1 @ Spsd)) @ u
Reference:
On optimal frequency-domain multichannel linear filtering
for noise reduction; M. Souden et al., 2010;
https://ieeexplore.ieee.org/document/5089420
Args:
psd_s (torch.complex64/ComplexTensor):
speech covariance matrix (..., F, C, C)
psd_n (torch.complex64/ComplexTensor):
observation/noise covariance matrix (..., F, C, C)
reference_vector (torch.Tensor): (..., C)
use_torch_solver (bool): Whether to use `solve` instead of `inverse`
diagonal_loading (bool): Whether to add a tiny term to the diagonal of psd_n
diag_eps (float):
eps (float):
Returns:
beamform_vector (torch.complex64/ComplexTensor): (..., F, C)
""" # noqa: D400
if diagonal_loading:
psd_n = tik_reg(psd_n, reg=diag_eps, eps=eps)
if use_torch_solver:
numerator = solve(psd_s, psd_n)
else:
numerator = matmul(inverse(psd_n), psd_s)
# NOTE (wangyou): until PyTorch 1.9.0, torch.trace does not
# support bacth processing. Use FC.trace() as fallback.
# ws: (..., C, C) / (...,) -> (..., C, C)
ws = numerator / (FC.trace(numerator)[..., None, None] + eps)
# h: (..., F, C_1, C_2) x (..., C_2) -> (..., F, C_1)
beamform_vector = einsum("...fec,...c->...fe", ws, reference_vector)
return beamform_vector
def get_WPD_filter_v2(
Phi: Union[torch.Tensor, ComplexTensor],
Rf: Union[torch.Tensor, ComplexTensor],
reference_vector: torch.Tensor,
diagonal_loading: bool = True,
diag_eps: float = 1e-7,
eps: float = 1e-8,
) -> Union[torch.Tensor, ComplexTensor]:
"""Return the WPD vector (v2).
This implementation is more efficient than `get_WPD_filter` as
it skips unnecessary computation with zeros.
Args:
Phi (torch.complex64/ComplexTensor): (B, F, C, C)
is speech PSD.
Rf (torch.complex64/ComplexTensor): (B, F, (btaps+1) * C, (btaps+1) * C)
is the power normalized spatio-temporal covariance matrix.
reference_vector (torch.Tensor): (B, C)
is the reference_vector.
diagonal_loading (bool):
Whether to add a tiny term to the diagonal of psd_n
diag_eps (float):
eps (float):
Returns:
filter_matrix (torch.complex64/ComplexTensor): (B, F, (btaps+1) * C)
"""
C = reference_vector.shape[-1]
if diagonal_loading:
Rf = tik_reg(Rf, reg=diag_eps, eps=eps)
inv_Rf = inverse(Rf)
# (B, F, (btaps+1) * C, C)
inv_Rf_pruned = inv_Rf[..., :C]
# numerator: (..., C_1, C_2) x (..., C_2, C_3) -> (..., C_1, C_3)
numerator = matmul(inv_Rf_pruned, Phi)
# NOTE (wangyou): until PyTorch 1.9.0, torch.trace does not
# support bacth processing. Use FC.trace() as fallback.
# ws: (..., (btaps+1) * C, C) / (...,) -> (..., (btaps+1) * C, C)
ws = numerator / (FC.trace(numerator[..., :C, :])[..., None, None] + eps)
# h: (..., F, C_1, C_2) x (..., C_2) -> (..., F, C_1)
beamform_vector = einsum("...fec,...c->...fe", ws, reference_vector)
# (B, F, (btaps+1) * C)
return beamform_vector
def test_matmul(real_vec):
if is_torch_1_9_plus:
wrappers = [ComplexTensor, torch.complex]
modules = [FC, torch]
else:
wrappers = [ComplexTensor]
modules = [FC]
for complex_wrapper, complex_module in zip(wrappers, modules):
mat = complex_wrapper(torch.rand(2, 3, 3), torch.rand(2, 3, 3))
if real_vec:
vec = torch.rand(2, 3, 1)
vec2 = complex_wrapper(vec, torch.zeros_like(vec))
else:
vec = complex_wrapper(torch.rand(2, 3, 1), torch.rand(2, 3, 1))
vec2 = vec
ret = matmul(mat, vec)
ret2 = complex_module.matmul(mat, vec2)
assert complex_module.allclose(ret, ret2)
def get_filter_matrix_conj(
correlation_matrix: Union[torch.Tensor, ComplexTensor],
correlation_vector: Union[torch.Tensor, ComplexTensor],
eps: float = 1e-10,
) -> Union[torch.Tensor, ComplexTensor]:
"""Calculate (conjugate) filter matrix based on correlations for one freq.
Args:
correlation_matrix : Correlation matrix (F, taps * C, taps * C)
correlation_vector : Correlation vector (F, taps, C, C)
eps:
Returns:
filter_matrix_conj (torch.complex/ComplexTensor): (F, taps, C, C)
"""
F, taps, C, _ = correlation_vector.size()
# (F, taps, C1, C2) -> (F, C1, taps, C2) -> (F, C1, taps * C2)
correlation_vector = (
correlation_vector.permute(0, 2, 1, 3).contiguous().view(F, C, taps * C)
)
eye = torch.eye(
correlation_matrix.size(-1),
dtype=correlation_matrix.dtype,
device=correlation_matrix.device,
)
shape = (
tuple(1 for _ in range(correlation_matrix.dim() - 2))
+ correlation_matrix.shape[-2:]
)
eye = eye.view(*shape)
correlation_matrix += eps * eye
inv_correlation_matrix = correlation_matrix.inverse()
# (F, C, taps, C) x (F, taps * C, taps * C) -> (F, C, taps * C)
stacked_filter_conj = matmul(
correlation_vector, inv_correlation_matrix.transpose(-1, -2)
)
# (F, C1, taps * C2) -> (F, C1, taps, C2) -> (F, taps, C2, C1)
filter_matrix_conj = stacked_filter_conj.view(F, C, taps, C).permute(0, 2, 3, 1)
return filter_matrix_conj
def get_mfmvdr_vector(gammax,
Phi,
use_torch_solver: bool = True,
eps: float = EPS):
"""Compute conventional MFMPDR/MFMVDR filter.
Args:
gammax (torch.complex64/ComplexTensor): (..., L, N)
Phi (torch.complex64/ComplexTensor): (..., L, N, N)
use_torch_solver (bool): Whether to use `solve` instead of `inverse`
eps (float)
Returns:
beamforming_vector (torch.complex64/ComplexTensor): (..., L, N)
"""
# (..., L, N)
if use_torch_solver:
numerator = solve(gammax.unsqueeze(-1), Phi).squeeze(-1)
else:
numerator = matmul(inverse(Phi), gammax.unsqueeze(-1)).squeeze(-1)
denominator = einsum("...d,...d->...", gammax.conj(), numerator)
return numerator / (denominator.real.unsqueeze(-1) + eps)
def get_mwf_vector(
psd_s,
psd_n,
reference_vector: Union[torch.Tensor, int],
use_torch_solver: bool = True,
diagonal_loading: bool = True,
diag_eps: float = 1e-7,
eps: float = 1e-8,
):
"""Return the MWF (Minimum Multi-channel Wiener Filter) vector:
h = (Npsd^-1 @ Spsd) @ u
Args:
psd_s (torch.complex64/ComplexTensor):
speech covariance matrix (..., F, C, C)
psd_n (torch.complex64/ComplexTensor):
power-normalized observation covariance matrix (..., F, C, C)
reference_vector (torch.Tensor or int): (..., C) or scalar
use_torch_solver (bool): Whether to use `solve` instead of `inverse`
diagonal_loading (bool): Whether to add a tiny term to the diagonal of psd_n
diag_eps (float):
eps (float):
Returns:
beamform_vector (torch.complex64/ComplexTensor): (..., F, C)
""" # noqa: D400
if diagonal_loading:
psd_n = tik_reg(psd_n, reg=diag_eps, eps=eps)
if use_torch_solver:
ws = solve(psd_s, psd_n)
else:
ws = matmul(inverse(psd_n), psd_s)
# h: (..., F, C_1, C_2) x (..., C_2) -> (..., F, C_1)
if isinstance(reference_vector, int):
beamform_vector = ws[..., reference_vector]
else:
beamform_vector = einsum("...fec,...c->...fe", ws, reference_vector)
return beamform_vector
def get_WPD_filter_with_rtf(
psd_observed_bar: Union[torch.Tensor, ComplexTensor],
psd_speech: Union[torch.Tensor, ComplexTensor],
psd_noise: Union[torch.Tensor, ComplexTensor],
iterations: int = 3,
reference_vector: Union[int, torch.Tensor, None] = None,
normalize_ref_channel: Optional[int] = None,
use_torch_solver: bool = True,
diagonal_loading: bool = True,
diag_eps: float = 1e-7,
eps: float = 1e-15,
) -> Union[torch.Tensor, ComplexTensor]:
"""Return the WPD vector calculated with RTF.
WPD is the Weighted Power minimization Distortionless response
convolutional beamformer. As follows:
h = (Rf^-1 @ vbar) / (vbar^H @ R^-1 @ vbar)
Reference:
T. Nakatani and K. Kinoshita, "A Unified Convolutional Beamformer
for Simultaneous Denoising and Dereverberation," in IEEE Signal
Processing Letters, vol. 26, no. 6, pp. 903-907, June 2019, doi:
10.1109/LSP.2019.2911179.
https://ieeexplore.ieee.org/document/8691481
Args:
psd_observed_bar (torch.complex64/ComplexTensor):
stacked observation covariance matrix
psd_speech (torch.complex64/ComplexTensor):
speech covariance matrix (..., F, C, C)
psd_noise (torch.complex64/ComplexTensor):
noise covariance matrix (..., F, C, C)
iterations (int): number of iterations in power method
reference_vector (torch.Tensor or int): (..., C) or scalar
normalize_ref_channel (int):
reference channel for normalizing the RTF
use_torch_solver (bool):
Whether to use `solve` instead of `inverse`
diagonal_loading (bool):
Whether to add a tiny term to the diagonal of psd_n
diag_eps (float):
eps (float):
Returns:
beamform_vector (torch.complex64/ComplexTensor)r: (..., F, C)
"""
if isinstance(psd_speech, ComplexTensor):
pad_func = FC.pad
elif is_torch_complex_tensor(psd_speech):
pad_func = torch.nn.functional.pad
else:
raise ValueError(
"Please update your PyTorch version to 1.9+ for complex support.")
C = psd_noise.size(-1)
if diagonal_loading:
psd_noise = tik_reg(psd_noise, reg=diag_eps, eps=eps)
# (B, F, C, 1)
rtf = get_rtf(
psd_speech,
psd_noise,
reference_vector,
iterations=iterations,
use_torch_solver=use_torch_solver,
)
# (B, F, (K+1)*C, 1)
rtf = pad_func(rtf, (0, 0, 0, psd_observed_bar.shape[-1] - C), "constant",
0)
# numerator: (..., C_1, C_2) x (..., C_2, 1) -> (..., C_1)
if use_torch_solver:
numerator = solve(rtf, psd_observed_bar).squeeze(-1)
else:
numerator = matmul(inverse(psd_observed_bar), rtf).squeeze(-1)
denominator = einsum("...d,...d->...", rtf.squeeze(-1).conj(), numerator)
if normalize_ref_channel is not None:
scale = rtf.squeeze(-1)[..., normalize_ref_channel, None].conj()
beamforming_vector = numerator * scale / (
denominator.real.unsqueeze(-1) + eps)
else:
beamforming_vector = numerator / (denominator.real.unsqueeze(-1) + eps)
return beamforming_vector
def get_mvdr_vector_with_rtf(
psd_n: Union[torch.Tensor, ComplexTensor],
psd_speech: Union[torch.Tensor, ComplexTensor],
psd_noise: Union[torch.Tensor, ComplexTensor],
iterations: int = 3,
reference_vector: Union[int, torch.Tensor, None] = None,
normalize_ref_channel: Optional[int] = None,
use_torch_solver: bool = True,
diagonal_loading: bool = True,
diag_eps: float = 1e-7,
eps: float = 1e-8,
) -> Union[torch.Tensor, ComplexTensor]:
"""Return the MVDR (Minimum Variance Distortionless Response) vector
calculated with RTF:
h = (Npsd^-1 @ rtf) / (rtf^H @ Npsd^-1 @ rtf)
Reference:
On optimal frequency-domain multichannel linear filtering
for noise reduction; M. Souden et al., 2010;
https://ieeexplore.ieee.org/document/5089420
Args:
psd_n (torch.complex64/ComplexTensor):
observation/noise covariance matrix (..., F, C, C)
psd_speech (torch.complex64/ComplexTensor):
speech covariance matrix (..., F, C, C)
psd_noise (torch.complex64/ComplexTensor):
noise covariance matrix (..., F, C, C)
iterations (int): number of iterations in power method
reference_vector (torch.Tensor or int): (..., C) or scalar
normalize_ref_channel (int): reference channel for normalizing the RTF
use_torch_solver (bool): Whether to use `solve` instead of `inverse`
diagonal_loading (bool): Whether to add a tiny term to the diagonal of psd_n
diag_eps (float):
eps (float):
Returns:
beamform_vector (torch.complex64/ComplexTensor): (..., F, C)
""" # noqa: H405, D205, D400
if diagonal_loading:
psd_noise = tik_reg(psd_noise, reg=diag_eps, eps=eps)
# (B, F, C, 1)
rtf = get_rtf(
psd_speech,
psd_noise,
reference_vector,
iterations=iterations,
use_torch_solver=use_torch_solver,
)
# numerator: (..., C_1, C_2) x (..., C_2, 1) -> (..., C_1)
if use_torch_solver:
numerator = solve(rtf, psd_n).squeeze(-1)
else:
numerator = matmul(inverse(psd_n), rtf).squeeze(-1)
denominator = einsum("...d,...d->...", rtf.squeeze(-1).conj(), numerator)
if normalize_ref_channel is not None:
scale = rtf.squeeze(-1)[..., normalize_ref_channel, None].conj()
beamforming_vector = numerator * scale / (
denominator.real.unsqueeze(-1) + eps)
else:
beamforming_vector = numerator / (denominator.real.unsqueeze(-1) + eps)
return beamforming_vector
def get_gev_vector(
psd_noise: Union[torch.Tensor, ComplexTensor],
psd_speech: Union[torch.Tensor, ComplexTensor],
mode="power",
reference_vector: Union[int, torch.Tensor] = 0,
iterations: int = 3,
use_torch_solver: bool = True,
diagonal_loading: bool = True,
diag_eps: float = 1e-7,
eps: float = 1e-8,
) -> Union[torch.Tensor, ComplexTensor]:
"""Return the generalized eigenvalue (GEV) beamformer vector:
psd_speech @ h = lambda * psd_noise @ h
Reference:
Blind acoustic beamforming based on generalized eigenvalue decomposition;
E. Warsitz and R. Haeb-Umbach, 2007.
Args:
psd_noise (torch.complex64/ComplexTensor):
noise covariance matrix (..., F, C, C)
psd_speech (torch.complex64/ComplexTensor):
speech covariance matrix (..., F, C, C)
mode (str): one of ("power", "evd")
"power": power method
"evd": eigenvalue decomposition (only for torch builtin complex tensors)
reference_vector (torch.Tensor or int): (..., C) or scalar
iterations (int): number of iterations in power method
use_torch_solver (bool): Whether to use `solve` instead of `inverse`
diagonal_loading (bool): Whether to add a tiny term to the diagonal of psd_n
diag_eps (float):
eps (float):
Returns:
beamform_vector (torch.complex64/ComplexTensor): (..., F, C)
""" # noqa: H405, D205, D400
if diagonal_loading:
psd_noise = tik_reg(psd_noise, reg=diag_eps, eps=eps)
if mode == "power":
if use_torch_solver:
phi = solve(psd_speech, psd_noise)
else:
phi = matmul(inverse(psd_noise), psd_speech)
e_vec = (phi[..., reference_vector,
None] if isinstance(reference_vector, int) else matmul(
phi, reference_vector[..., None, :, None]))
for _ in range(iterations - 1):
e_vec = matmul(phi, e_vec)
# e_vec = e_vec / complex_norm(e_vec, dim=-1, keepdim=True)
e_vec = e_vec.squeeze(-1)
elif mode == "evd":
assert (is_torch_1_9_plus and is_torch_complex_tensor(psd_speech)
and is_torch_complex_tensor(psd_noise))
# e_vec = generalized_eigenvalue_decomposition(psd_speech, psd_noise)[1][...,-1]
e_vec = psd_noise.new_zeros(psd_noise.shape[:-1])
for f in range(psd_noise.shape[-3]):
try:
e_vec[..., f, :] = generalized_eigenvalue_decomposition(
psd_speech[..., f, :, :], psd_noise[..., f, :, :])[1][...,
-1]
except RuntimeError:
# port from github.com/fgnt/nn-gev/blob/master/fgnt/beamforming.py#L106
print(
"GEV beamformer: LinAlg error for frequency {}".format(f),
flush=True,
)
C = psd_noise.size(-1)
e_vec[...,
f, :] = (psd_noise.new_ones(e_vec[..., f, :].shape) /
FC.trace(psd_noise[..., f, :, :]) * C)
else:
raise ValueError("Unknown mode: %s" % mode)
beamforming_vector = e_vec / complex_norm(e_vec, dim=-1, keepdim=True)
beamforming_vector = gev_phase_correction(beamforming_vector)
return beamforming_vector
def get_rank1_mwf_vector(
psd_speech,
psd_noise,
reference_vector: Union[torch.Tensor, int],
denoising_weight: float = 1.0,
approx_low_rank_psd_speech: bool = False,
iterations: int = 3,
use_torch_solver: bool = True,
diagonal_loading: bool = True,
diag_eps: float = 1e-7,
eps: float = 1e-8,
):
"""Return the R1-MWF (Rank-1 Multi-channel Wiener Filter) vector
h = (Npsd^-1 @ Spsd) / (mu + Tr(Npsd^-1 @ Spsd)) @ u
Reference:
[1] Rank-1 constrained multichannel Wiener filter for speech recognition in
noisy environments; Z. Wang et al, 2018
https://hal.inria.fr/hal-01634449/document
[2] Low-rank approximation based multichannel Wiener filter algorithms for
noise reduction with application in cochlear implants; R. Serizel, 2014
https://ieeexplore.ieee.org/document/6730918
Args:
psd_speech (torch.complex64/ComplexTensor):
speech covariance matrix (..., F, C, C)
psd_noise (torch.complex64/ComplexTensor):
noise covariance matrix (..., F, C, C)
reference_vector (torch.Tensor or int): (..., C) or scalar
denoising_weight (float): a trade-off parameter between noise reduction and
speech distortion.
A larger value leads to more noise reduction at the expense of more speech
distortion.
When `denoising_weight = 0`, it corresponds to MVDR beamformer.
approx_low_rank_psd_speech (bool): whether to replace original input psd_speech
with its low-rank approximation as in [1]
iterations (int): number of iterations in power method, only used when
`approx_low_rank_psd_speech = True`
use_torch_solver (bool): Whether to use `solve` instead of `inverse`
diagonal_loading (bool): Whether to add a tiny term to the diagonal of psd_n
diag_eps (float):
eps (float):
Returns:
beamform_vector (torch.complex64/ComplexTensor): (..., F, C)
""" # noqa: H405, D205, D400
if approx_low_rank_psd_speech:
if diagonal_loading:
psd_noise = tik_reg(psd_noise, reg=diag_eps, eps=eps)
# (B, F, C, 1)
recon_vec = get_rtf(
psd_speech,
psd_noise,
mode="power",
iterations=iterations,
reference_vector=reference_vector,
use_torch_solver=use_torch_solver,
)
# Eq. (25) in Ref[1]
psd_speech_r1 = matmul(recon_vec, recon_vec.conj().transpose(-1, -2))
sigma_speech = FC.trace(psd_speech) / (FC.trace(psd_speech_r1) + eps)
psd_speech_r1 = psd_speech_r1 * sigma_speech[..., None, None]
# c.f. Eq. (62) in Ref[2]
psd_speech = psd_speech_r1
elif diagonal_loading:
psd_noise = tik_reg(psd_noise, reg=diag_eps, eps=eps)
if use_torch_solver:
numerator = solve(psd_speech, psd_noise)
else:
numerator = matmul(inverse(psd_noise), psd_speech)
# NOTE (wangyou): until PyTorch 1.9.0, torch.trace does not
# support bacth processing. Use FC.trace() as fallback.
# ws: (..., C, C) / (...,) -> (..., C, C)
ws = numerator / (denoising_weight + FC.trace(numerator)[..., None, None] +
eps)
# h: (..., F, C_1, C_2) x (..., C_2) -> (..., F, C_1)
if isinstance(reference_vector, int):
beamform_vector = ws[..., reference_vector]
else:
beamform_vector = einsum("...fec,...c->...fe", ws, reference_vector)
return beamform_vector
def get_sdw_mwf_vector(
psd_speech,
psd_noise,
reference_vector: Union[torch.Tensor, int],
denoising_weight: float = 1.0,
approx_low_rank_psd_speech: bool = False,
iterations: int = 3,
use_torch_solver: bool = True,
diagonal_loading: bool = True,
diag_eps: float = 1e-7,
eps: float = 1e-8,
):
"""Return the SDW-MWF (Speech Distortion Weighted Multi-channel Wiener Filter) vector
h = (Spsd + mu * Npsd)^-1 @ Spsd @ u
Reference:
[1] Spatially pre-processed speech distortion weighted multi-channel Wiener
filtering for noise reduction; A. Spriet et al, 2004
https://dl.acm.org/doi/abs/10.1016/j.sigpro.2004.07.028
[2] Rank-1 constrained multichannel Wiener filter for speech recognition in
noisy environments; Z. Wang et al, 2018
https://hal.inria.fr/hal-01634449/document
[3] Low-rank approximation based multichannel Wiener filter algorithms for
noise reduction with application in cochlear implants; R. Serizel, 2014
https://ieeexplore.ieee.org/document/6730918
Args:
psd_speech (torch.complex64/ComplexTensor):
speech covariance matrix (..., F, C, C)
psd_noise (torch.complex64/ComplexTensor):
noise covariance matrix (..., F, C, C)
reference_vector (torch.Tensor or int): (..., C) or scalar
denoising_weight (float): a trade-off parameter between noise reduction and
speech distortion.
A larger value leads to more noise reduction at the expense of more speech
distortion.
The plain MWF is obtained with `denoising_weight = 1` (by default).
approx_low_rank_psd_speech (bool): whether to replace original input psd_speech
with its low-rank approximation as in [2]
iterations (int): number of iterations in power method, only used when
`approx_low_rank_psd_speech = True`
use_torch_solver (bool): Whether to use `solve` instead of `inverse`
diagonal_loading (bool): Whether to add a tiny term to the diagonal of psd_n
diag_eps (float):
eps (float):
Returns:
beamform_vector (torch.complex64/ComplexTensor): (..., F, C)
""" # noqa: H405, D205, D400
if approx_low_rank_psd_speech:
if diagonal_loading:
psd_noise = tik_reg(psd_noise, reg=diag_eps, eps=eps)
# (B, F, C, 1)
recon_vec = get_rtf(
psd_speech,
psd_noise,
mode="power",
iterations=iterations,
reference_vector=reference_vector,
use_torch_solver=use_torch_solver,
)
# Eq. (25) in Ref[2]
psd_speech_r1 = matmul(recon_vec, recon_vec.conj().transpose(-1, -2))
sigma_speech = FC.trace(psd_speech) / (FC.trace(psd_speech_r1) + eps)
psd_speech_r1 = psd_speech_r1 * sigma_speech[..., None, None]
# c.f. Eq. (62) in Ref[3]
psd_speech = psd_speech_r1
psd_n = psd_speech + denoising_weight * psd_noise
if diagonal_loading:
psd_n = tik_reg(psd_n, reg=diag_eps, eps=eps)
if use_torch_solver:
ws = solve(psd_speech, psd_n)
else:
ws = matmul(inverse(psd_n), psd_speech)
if isinstance(reference_vector, int):
beamform_vector = ws[..., reference_vector]
else:
beamform_vector = einsum("...fec,...c->...fe", ws, reference_vector)
return beamform_vector