def fft_layer(self, data_x):
# Creating FFT
filter_num = 100
data_input = data_x
data_x_padded = keras.layers.ZeroPadding1D((0, 1))(data_input)
data_x_squeez = keras.layers.Lambda(
lambda x: keras.backend.squeeze(x, 2),
output_shape=lambda shape: (shape[0], shape[1]),
name='data_x_squeez')(data_x_padded)
data_x_rfft = keras.layers.Lambda(lambda x: tf.signal.rfft(x),
name='data_x_rfft')(data_x_squeez)
data_x_rfft_abs = keras.layers.Lambda(
lambda x: tf.math.abs(x), name='data_x_rfft_abs')(data_x_rfft)
data_x_rfft_angle_complex = keras.layers.Lambda(
lambda x: tf.math.exp(tf.complex(0.0, tf.math.angle(x))),
name='data_x_rfft_angle_complex')(data_x_rfft)
# Learning Filters 800*100 100
rfft_process = data_x_rfft_abs
rfft_process_dense = keras.layers.Dense(
units=filter_num,
activation="sigmoid",
use_bias=False,
kernel_initializer=keras.initializers.Identity())(rfft_process)
# rfft_process_dense = layers.AddSingletonDepth()(rfft_process_dense)
rfft_process_dense = keras.layers.RepeatVector(
rfft_process.shape[-1])(rfft_process_dense)
rfft_process = layers.AddSingletonDepth()(data_x_rfft_abs)
rfft_process_filter = layers.BigFilter(
units=filter_num, name="freq_filters")(rfft_process)
rfft_process = keras.layers.Multiply(name='rfft_process')(
[rfft_process_dense, rfft_process_filter])
rfft_process = keras.layers.Convolution1D(1, 1)(rfft_process)
# rfft_process = self.activation(rfft_process)
#Recontruction
rfft_process_complex = keras.layers.Lambda(
lambda x: tf.complex(tf.math.abs(x), 0.0),
name='rfft_process_complex')(rfft_process)
new_rfft = keras.layers.Multiply(name='new_rfft')([
layers.AddSingletonDepth()(data_x_rfft_angle_complex),
rfft_process_complex
])
new_rfft = keras.layers.Lambda(lambda x: keras.backend.squeeze(x, -1))(
new_rfft)
data_out = keras.layers.Lambda(lambda x: tf.signal.irfft(x),
name='IRFFT')(new_rfft)
data_out = keras.layers.Lambda(
lambda x: keras.backend.expand_dims(x, -1))(data_out)
data_out = tf.keras.layers.Cropping1D(cropping=(0, 1),
name="data_fft_put")(data_out)
return data_out
def build_model(self):
data_input = keras.engine.Input(
shape=(self.input_length,),
name='data_input')
data_expanded = layers.AddSingletonDepth()(data_input)
data_out = keras.layers.Convolution1D(self.config['model']['filters']['depths']['res'],
self.config['model']['filters']['lengths']['res'], padding='same',
use_bias=False,
name='initial_causal_conv')(data_expanded)
skip_connections = []
res_block_i = 0
for stack_i in range(self.num_stacks):
layer_in_stack = 0
for dilation in self.dilations:
res_block_i += 1
data_out, skip_out = self.dilated_residual_block(data_out, res_block_i, layer_in_stack, dilation, stack_i)
if skip_out is not None:
skip_connections.append(skip_out)
layer_in_stack += 1
data_out = keras.layers.Add()(skip_connections)
data_out = self.activation(data_out)
data_out = keras.layers.Convolution1D(self.config['model']['filters']['depths']['final'][0],
self.config['model']['filters']['lengths']['final'][0],
padding='same',
use_bias=False)(data_out)
data_out = self.activation(data_out)
data_out = keras.layers.Convolution1D(self.config['model']['filters']['depths']['final'][1],
self.config['model']['filters']['lengths']['final'][1], padding='same',
use_bias=False)(data_out)
data_out = keras.layers.Convolution1D(3, 1)(data_out)
data_out_vocals = layers.Slice((Ellipsis, slice(0, 1)), (self.padded_target_field_length, 1),
name='slice_data_output_1')(data_out)
data_out_drums = layers.Slice((Ellipsis, slice(1, 2)), (self.padded_target_field_length, 1),
name='slice_data_output_2')(data_out)
data_out_bass = layers.Slice((Ellipsis, slice(2, 3)), (self.padded_target_field_length, 1),
name='slice_data_output_3')(data_out)
data_out_vocals = keras.layers.Lambda(lambda x: keras.backend.squeeze(x, 2),
output_shape=lambda shape: (shape[0], shape[1]), name='data_output_1')(
data_out_vocals)
data_out_drums = keras.layers.Lambda(lambda x: keras.backend.squeeze(x, 2),
output_shape=lambda shape: (shape[0], shape[1]), name='data_output_2')(
data_out_drums)
data_out_bass = keras.layers.Lambda(lambda x: keras.backend.squeeze(x, 2),
output_shape=lambda shape: (shape[0], shape[1]), name='data_output_3')(
data_out_bass)
return keras.engine.Model(inputs=[data_input], outputs=[data_out_vocals, data_out_drums, data_out_bass])
def build_model(self):
data_input = keras.engine.Input(
shape=(self.input_length,),
name='data_input')
condition_input = keras.engine.Input(shape=(self.condition_input_length,),
name='condition_input')
data_expanded = layers.AddSingletonDepth()(data_input)
data_input_target_field_length = layers.Slice(
(slice(self.samples_of_interest_indices[0], self.samples_of_interest_indices[-1] + 1, 1), Ellipsis),
(self.padded_target_field_length,1),
name='data_input_target_field_length')(data_expanded)
data_out = keras.layers.Conv1D(self.config['model']['filters']['depths']['res'],
self.config['model']['filters']['lengths']['res'], padding='same',
use_bias=False,
name='initial_causal_conv')(data_expanded)
condition_out = keras.layers.Dense(self.config['model']['filters']['depths']['res'],
name='initial_dense_condition',
use_bias=False)(condition_input)
condition_out = keras.layers.RepeatVector(int(self.input_length),
name='initial_condition_repeat')(condition_out)
data_out = keras.layers.Add(name='initial_data_condition_merge')([data_out, condition_out])
skip_connections = []
res_block_i = 0
for stack_i in range(self.num_stacks):
layer_in_stack = 0
for dilation in self.dilations:
res_block_i += 1
data_out, skip_out = self.dilated_residual_block(data_out, condition_input, res_block_i, layer_in_stack, dilation, stack_i)
if skip_out is not None:
skip_connections.append(skip_out)
layer_in_stack += 1
data_out = keras.layers.Add()(skip_connections)
data_out = self.activation(data_out)
data_out = keras.layers.Conv1D(self.config['model']['filters']['depths']['final'][0],
self.config['model']['filters']['lengths']['final'][0],
padding='same',
bias=False)(data_out)
condition_out = keras.layers.Dense(self.config['model']['filters']['depths']['final'][0],
bias=False,
name='penultimate_conv_1d_condition')(condition_input)
condition_out = keras.layers.RepeatVector(self.padded_target_field_length,
name='penultimate_conv_1d_condition_repeat')(condition_out)
data_out = keras.layers.Add(name='penultimate_conv_1d_condition_merge')([data_out, condition_out])
data_out = self.activation(data_out)
data_out = keras.layers.Conv1D(self.config['model']['filters']['depths']['final'][1],
self.config['model']['filters']['lengths']['final'][1], padding='same',
use_bias=False)(data_out)
condition_out = keras.layers.Dense(self.config['model']['filters']['depths']['final'][1], use_bias=False,
name='final_conv_1d_condition')(condition_input)
condition_out = keras.layers.RepeatVector(self.padded_target_field_length,
name='final_conv_1d_condition_repeat')(condition_out)
data_out = keras.layers.Add(name='final_conv_1d_condition_merge')([data_out, condition_out])
data_out = keras.layers.Conv1D(1, 1)(data_out)
data_out_speech = data_out
data_out_noise = layers.Subtract(name='subtract_layer')([data_input_target_field_length, data_out_speech])
data_out_speech = keras.layers.Lambda(lambda x: keras.backend.squeeze(x, 2),
output_shape=lambda shape: (shape[0], shape[1]), name='data_output_1')(
data_out_speech)
data_out_noise = keras.layers.Lambda(lambda x: keras.backend.squeeze(x, 2),
output_shape=lambda shape: (shape[0], shape[1]), name='data_output_2')(
data_out_noise)
return keras.engine.Model(input=[data_input, condition_input], output=[data_out_speech, data_out_noise])
def build_model(self):
data_input = keras.layers.Input(shape=(
2,
self.input_length,
),
name='data_input')
# condition_input = keras.engine.Input(shape=(self.condition_input_length,),
# name='condition_input')
data_mix = keras.layers.Lambda(lambda x: keras.backend.sum(x, 1))(
data_input)
data_expanded = layers.AddSingletonDepth()(data_mix)
# data_input_target_field_length = layers.Slice(
# (slice(self.samples_of_interest_indices[0], self.samples_of_interest_indices[-1] + 1, 1), Ellipsis),
# (self.padded_target_field_length,1),
# name='data_input_target_field_length')(data_expanded)
data_out = keras.layers.Conv1D(
self.config['model']['filters']['depths']['res'],
self.config['model']['filters']['lengths']['res'],
padding='same',
use_bias=False,
name='initial_causal_conv')(data_expanded)
# condition_out = keras.layers.Dense(self.config['model']['filters']['depths']['res'],
# name='initial_dense_condition',
# bias=False)(condition_input)
# condition_out = keras.layers.RepeatVector(self.input_length,
# name='initial_condition_repeat')(condition_out)
# data_out = keras.layers.Merge(mode='sum', name='initial_data_condition_merge')(
# [data_out, condition_out])
skip_connections = []
res_block_i = 0
for stack_i in range(self.num_stacks):
layer_in_stack = 0
for dilation in self.dilations:
res_block_i += 1
data_out, skip_out = self.dilated_residual_block(data_out, res_block_i, \
layer_in_stack, dilation, stack_i)
if skip_out is not None:
skip_connections.append(skip_out)
layer_in_stack += 1
data_out = keras.layers.Add()(skip_connections)
data_out = self.activation(data_out)
data_out = keras.layers.Conv1D(
self.config['model']['filters']['depths']['final'][0],
self.config['model']['filters']['lengths']['final'][0],
padding='same',
use_bias=False)(data_out)
# condition_out = keras.layers.Dense(self.config['model']['filters']['depths']['final'][0],
# bias=False,
# name='penultimate_conv_1d_condition')(condition_input)
# condition_out = keras.layers.RepeatVector(self.padded_target_field_length,
# name='penultimate_conv_1d_condition_repeat')(condition_out)
# data_out = keras.layers.Merge(mode='sum', name='penultimate_conv_1d_condition_merge')([data_out, condition_out])
data_out = self.activation(data_out)
data_out = keras.layers.Conv1D(
self.config['model']['filters']['depths']['final'][1],
self.config['model']['filters']['lengths']['final'][1],
padding='same',
use_bias=False)(data_out)
# condition_out = keras.layers.Dense(self.config['model']['filters']['depths']['final'][1], bias=False,
# name='final_conv_1d_condition')(condition_input)
# condition_out = keras.layers.RepeatVector(self.padded_target_field_length,
# name='final_conv_1d_condition_repeat')(condition_out)
# data_out = keras.layers.Merge(mode='sum', name='final_conv_1d_condition_merge')([data_out, condition_out])
# data_out = keras.layers.Conv1D(self.n_output, 1)(data_out)
data_out = keras.layers.Conv1D(
self.n_output,
self.config['model']['filters']['lengths']['final'][1],
padding='same',
use_bias=False)(data_out)
out_speech = keras.layers.Lambda(
lambda x: keras.backend.permute_dimensions(x, (0, 2, 1)),
name='data_output')(data_out)
# out_speech_2 = keras.layers.Lambda(lambda x: x[:,:,1],
# output_shape=lambda shape: (shape[0], shape[1]),
# name='data_output_2')(data_out)
## add discriminator
return keras.engine.Model(inputs=[data_input], outputs=[out_speech])
def build_model(self):
data_input = keras.Input(shape=(self.input_length, ),
name='data_input')
data_expanded = layers.AddSingletonDepth()(data_input)
data_input_target_field_length = layers.Slice(
(slice(self.samples_of_interest_indices[0],
self.samples_of_interest_indices[-1] + 1, 1), Ellipsis),
(self.padded_target_field_length, 1),
name='data_input_target_field_length')(data_expanded)
data_out = keras.layers.Convolution1D(
self.config['model']['filters']['depths']['res'],
self.config['model']['filters']['lengths']['res'],
padding='same',
use_bias=False,
name='initial_causal_conv_1',
kernel_constraint=keras.constraints.max_norm(30))(data_expanded)
skip_connections = []
res_block_i = 0
for stack_i in range(self.num_stacks):
layer_in_stack = 0
for dilation in self.dilations:
res_block_i += 1
data_out, skip_out = self.dilated_residual_block(
data_out, res_block_i, layer_in_stack, dilation, stack_i)
if skip_out is not None:
skip_connections.append(skip_out)
layer_in_stack += 1
data_out = keras.layers.Add()(skip_connections)
# data_out = keras.layers.Concatenate(axis=2)(skip_connections)
# data_out = keras.layers.Convolution1D(self.config['model']['filters']['depths']['res'], 1)(data_out)
data_out = self.activation(data_out)
data_out = keras.layers.Convolution1D(
self.config['model']['filters']['depths']['final'][0],
self.config['model']['filters']['lengths']['final'][0],
padding='same',
use_bias=False)(data_out)
data_out = self.activation(data_out)
data_out = keras.layers.Convolution1D(
self.config['model']['filters']['depths']['final'][1],
self.config['model']['filters']['lengths']['final'][1],
padding='same',
use_bias=False)(data_out)
data_out = keras.layers.Convolution1D(1, 1)(data_out)
data_out_speech = self.fft_layer(data_out)
data_out_noise = layers.Subtract(name='subtract_layer')(
[data_input_target_field_length, data_out_speech])
data_out_speech = keras.layers.Lambda(
lambda x: keras.backend.squeeze(x, -1),
output_shape=lambda shape: (shape[0], shape[1]),
name='data_output_1')(data_out_speech)
data_out_noise = keras.layers.Lambda(
lambda x: keras.backend.squeeze(x, -1),
output_shape=lambda shape: (shape[0], shape[1]),
name='data_output_2')(data_out_noise)
return keras.Model(inputs=[data_input],
outputs=[data_out_speech, data_out_noise])