def input_target_spec(s, d, s_len, d_len, SNR, N_w, N_s, NFFT, f_s):
'''
Input features and target (spectrum) for polar form acoustic-domain.
Inputs:
s - clean speech (dtype=tf.int32).
d - noise (dtype=tf.int32).
s_len - clean speech length without padding (samples).
d_len - noise length without padding (samples).
SNR - SNR level.
N_w - time-domain window length (samples).
N_s - time-domain window shift (samples).
NFFT - number of acoustic-domain DFT components.
f_s - sampling frequency (Hz).
Outputs:
x_MAG - noisy speech magnitude spectrum.
s_MAG - clean speech magnitude spectrum (target).
L - number of time-domain frames for each sequence.
'''
(x, s, _) = add_noise_batch(s, d, s_len, d_len, SNR)
L = num_frames(
s_len,
N_s) # number of time-domain frames for each sequence (uppercase eta).
x_MAG, _ = polar.analysis(x, N_w, N_s, NFFT)
s_MAG, _ = polar.analysis(s, N_w, N_s, NFFT)
s_MAG = tf.boolean_mask(s_MAG, tf.sequence_mask(L))
return x_MAG, s_MAG, L
def target_xi(s, d, s_len, d_len, SNR, N_w, N_s, NFFT, f_s):
'''
Target (a priori SNR) for polar form acoustic-domain.
Inputs:
s - clean speech (dtype=tf.int32).
d - noise (dtype=tf.int32).
s_len - clean speech length without padding (samples).
d_len - noise length without padding (samples).
SNR - SNR level.
N_w - time-domain window length (samples).
N_s - time-domain window shift (samples).
NFFT - number of acoustic-domain DFT components.
f_s - sampling frequency (Hz).
Outputs:
xi_dB - a priori SNR in dB (target).
L - number of time-domain frames for each sequence.
'''
(_, s, d) = add_noise_batch(s, d, s_len, d_len, SNR)
L = num_frames(
s_len, N_s
) # number of acoustic-domain frames for each sequence (uppercase eta).
s_MAG, _ = polar.analysis(s, N_w, N_s, NFFT)
d_MAG, _ = polar.analysis(d, N_w, N_s, NFFT)
s_MAG = tf.boolean_mask(s_MAG, tf.sequence_mask(L))
d_MAG = tf.boolean_mask(d_MAG, tf.sequence_mask(L))
xi = tf.truediv(tf.square(tf.maximum(s_MAG, 1e-12)),
tf.square(tf.maximum(d_MAG, 1e-12))) # a priori SNR.
xi_dB = tf.multiply(10.0, log10(xi)) # a priori SNR in dB.
return xi_dB, L
def input_target_xi(s, d, s_len, d_len, SNR, N_w, N_s, NFFT, f_s, mu, sigma):
'''
Input features and target (mapped a priori SNR) for polar form acoustic-domain.
Inputs:
s - clean speech (dtype=tf.int32).
d - noise (dtype=tf.int32).
s_len - clean speech length without padding (samples).
d_len - noise length without padding (samples).
SNR - SNR level.
N_w - time-domain window length (samples).
N_s - time-domain window shift (samples).
NFFT - number of acoustic-domain DFT components.
f_s - sampling frequency (Hz).
mu - sample mean.
sigma - sample standard deviation.
Outputs:
x_MAG - noisy speech magnitude spectrum.
xi_mapped - mapped a priori SNR (target).
L - number of time-domain frames for each sequence.
'''
(x, s, d) = add_noise_batch(s, d, s_len, d_len, SNR)
L = num_frames(
s_len, N_s
) # number of acoustic-domain frames for each sequence (uppercase eta).
x_MAG, _ = polar.analysis(x, N_w, N_s, NFFT)
s_MAG, _ = polar.analysis(s, N_w, N_s, NFFT)
s_MAG = tf.boolean_mask(s_MAG, tf.sequence_mask(L))
d_MAG, _ = polar.analysis(d, N_w, N_s, NFFT)
d_MAG = tf.boolean_mask(d_MAG, tf.sequence_mask(L))
xi_bar = xi.xi_bar(s_MAG, d_MAG, mu, sigma)
return x_MAG, xi_bar, L
def input_target_xi(s, d, s_len, d_len, SNR, N_w, N_s, NFFT, f_s, mu, sigma):
'''
Input features and target (mapped a priori SNR) for polar form acoustic-domain.
Inputs:
s - clean speech (dtype=tf.int32).
d - noise (dtype=tf.int32).
s_len - clean speech length without padding (samples).
d_len - noise length without padding (samples).
SNR - SNR level.
N_w - time-domain window length (samples).
N_s - time-domain window shift (samples).
NFFT - number of acoustic-domain DFT components.
f_s - sampling frequency (Hz).
mu - sample mean.
sigma - sample standard deviation.
Outputs:
x_MAG - noisy speech magnitude spectrum.
xi_mapped - mapped a priori SNR (target).
L - number of time-domain frames for each sequence.
'''
(x, s, d) = add_noise_batch(s, d, s_len, d_len, SNR)
L = num_frames(
s_len, N_s
) # number of acoustic-domain frames for each sequence (uppercase eta).
x_MAG, _ = polar.analysis(x, N_w, N_s, NFFT)
s_MAG, _ = polar.analysis(s, N_w, N_s, NFFT)
s_MAG = tf.boolean_mask(s_MAG, tf.sequence_mask(L))
d_MAG, _ = polar.analysis(d, N_w, N_s, NFFT)
d_MAG = tf.boolean_mask(d_MAG, tf.sequence_mask(L))
xi = tf.truediv(tf.square(tf.maximum(s_MAG, 1e-12)),
tf.square(tf.maximum(d_MAG, 1e-12))) # a priori SNR.
xi_dB = tf.multiply(10.0, log10(xi)) # a priori SNR in dB.
xi_mapped = tf.multiply(
0.5,
tf.add(
1.0,
tf.erf(
tf.truediv(tf.subtract(xi_dB, mu),
tf.multiply(
sigma, tf.sqrt(2.0)))))) # mapped a priori SNR.
return x_MAG, xi_mapped, L
def input(z, z_len, N_w, N_s, NFFT, f_s):
'''
Input features for polar form acoustic-domain.
Input/s:
z - speech (dtype=tf.int32).
z_len - speech length without padding (samples).
N_w - time-domain window length (samples).
N_s - time-domain window shift (samples).
NFFT - number of acoustic-domain DFT components.
f_s - sampling frequency (Hz).
Output/s:
z_MAG - speech magnitude spectrum.
z_PHA - speech phase spectrum.
L - number of time-domain frames for each sequence.
'''
z = tf.truediv(tf.cast(z, tf.float32), 32768.0)
L = num_frames(z_len, N_s)
z_MAG, z_PHA = polar.analysis(z, N_w, N_s, NFFT)
return z_MAG, L, z_PHA