def __init__(self, sample_rate, window_size, hop_size, mel_bins, fmin,
fmax, classes_num):
super(Cnn_9layers_FrameAtt, self).__init__()
window = 'hann'
center = True
pad_mode = 'reflect'
ref = 1.0
amin = 1e-10
top_db = None
# Spectrogram extractor
self.spectrogram_extractor = Spectrogram(n_fft=window_size, hop_length=hop_size,
win_length=window_size, window=window, center=center, pad_mode=pad_mode,
freeze_parameters=True)
# Logmel feature extractor
self.logmel_extractor = LogmelFilterBank(sr=sample_rate, n_fft=window_size,
n_mels=mel_bins, fmin=fmin, fmax=fmax, ref=ref, amin=amin, top_db=top_db,
freeze_parameters=True)
# Spec augmenter
self.spec_augmenter = SpecAugmentation(time_drop_width=64, time_stripes_num=2,
freq_drop_width=8, freq_stripes_num=2)
self.bn0 = nn.BatchNorm2d(64)
self.conv_block1 = ConvBlock(in_channels=1, out_channels=64)
self.conv_block2 = ConvBlock(in_channels=64, out_channels=128)
self.conv_block3 = ConvBlock(in_channels=128, out_channels=256)
self.conv_block4 = ConvBlock(in_channels=256, out_channels=512)
self.att_block = AttBlock(n_in=512, n_out=17, activation='sigmoid')
self.init_weights()
def __init__(self, sample_rate, window_size, hop_size, mel_bins, fmin,
fmax, classes_num):
super(Cnn_9layers_Gru_FrameAvg, self).__init__()
window = 'hann'
center = True
pad_mode = 'reflect'
ref = 1.0
amin = 1e-10
top_db = None
# Spectrogram extractor
self.spectrogram_extractor = Spectrogram(n_fft=window_size,
hop_length=hop_size,
win_length=window_size,
window=window,
center=center,
pad_mode=pad_mode,
freeze_parameters=True)
# Logmel feature extractor
self.logmel_extractor = LogmelFilterBank(sr=sample_rate,
n_fft=window_size,
n_mels=mel_bins,
fmin=fmin,
fmax=fmax,
ref=ref,
amin=amin,
top_db=top_db,
freeze_parameters=True)
# Spec augmenter
self.spec_augmenter = SpecAugmentation(time_drop_width=64,
time_stripes_num=2,
freq_drop_width=8,
freq_stripes_num=2)
self.bn0 = nn.BatchNorm2d(64)
self.conv_block1 = ConvBlock(in_channels=1, out_channels=64)
self.conv_block2 = ConvBlock(in_channels=64, out_channels=128)
self.conv_block3 = ConvBlock(in_channels=128, out_channels=256)
self.conv_block4 = ConvBlock(in_channels=256, out_channels=512)
self.gru = nn.GRU(input_size=512,
hidden_size=256,
num_layers=1,
bias=True,
batch_first=True,
bidirectional=True)
self.fc = nn.Linear(512, classes_num, bias=True)
self.init_weights()
def __init__(self, sample_rate, window_size, hop_size, mel_bins, fmin,
fmax, classes_num):
super(Cnn_9layers_Transformer_FrameAvg, self).__init__()
window = 'hann'
center = True
pad_mode = 'reflect'
ref = 1.0
amin = 1e-10
top_db = None
# Spectrogram extractor
self.spectrogram_extractor = Spectrogram(n_fft=window_size, hop_length=hop_size,
win_length=window_size, window=window, center=center, pad_mode=pad_mode,
freeze_parameters=True)
# Logmel feature extractor
self.logmel_extractor = LogmelFilterBank(sr=sample_rate, n_fft=window_size,
n_mels=mel_bins, fmin=fmin, fmax=fmax, ref=ref, amin=amin, top_db=top_db,
freeze_parameters=True)
# Spec augmenter
self.spec_augmenter = SpecAugmentation(time_drop_width=64, time_stripes_num=2,
freq_drop_width=8, freq_stripes_num=2)
self.bn0 = nn.BatchNorm2d(64)
self.conv_block1 = ConvBlock(in_channels=1, out_channels=64)
self.conv_block2 = ConvBlock(in_channels=64, out_channels=128)
self.conv_block3 = ConvBlock(in_channels=128, out_channels=256)
self.conv_block4 = ConvBlock(in_channels=256, out_channels=512)
n_head = 8
n_hid = 512
d_k = 64
d_v = 64
dropout = 0.2
self.multihead = MultiHead(n_head, n_hid, d_k, d_v, dropout)
self.fc = nn.Linear(512, classes_num, bias=True)
self.init_weights()
def __init__(self, sample_rate, window_size, hop_size, mel_bins, fmin,
fmax, classes_num):
super(Cnn14, self).__init__()
window = 'hann'
center = True
pad_mode = 'reflect'
ref = 1.0
amin = 1e-10
top_db = None
# Spectrogram extractor
self.spectrogram_extractor_openmic = Spectrogram(n_fft=window_size, hop_length=hop_size,
win_length=window_size, window=window, center=center, pad_mode=pad_mode,
freeze_parameters=True)
# Logmel feature extractor
self.logmel_extractor_openmic = LogmelFilterBank(sr=sample_rate, n_fft=window_size,
n_mels=mel_bins, fmin=fmin, fmax=fmax, ref=ref, amin=amin, top_db=top_db,
freeze_parameters=True)
self.bn0 = nn.BatchNorm2d(64)
self.conv_block1 = ConvBlock(in_channels=1, out_channels=64)
self.conv_block2 = ConvBlock(in_channels=64, out_channels=128)
self.conv_block3 = ConvBlock(in_channels=128, out_channels=256)
self.conv_block4 = ConvBlock(in_channels=256, out_channels=512)
# self.conv_block5 = ConvBlock(in_channels=512, out_channels=1024)
# self.conv_block6 = ConvBlock(in_channels=1024, out_channels=2048)
# self.fc1 = nn.Linear(2048, 2048, bias=True)
# self.fc_openmic = nn.Linear(2048, classes_num, bias=True)
self.fc1 = nn.Linear(512, 512, bias=True)
self.fc_monophonic = nn.Linear(512, classes_num, bias=True)
self.init_weight()
n_fft=window_size,
hop_length=hop_size,
win_length=window_size,
window=window,
center=center,
pad_mode=pad_mode,
freeze_parameters=True,
)
# Logmel feature extractor
logmel_extractor_openmic = LogmelFilterBank(
sr=sample_rate,
n_fft=window_size,
n_mels=mel_bins,
fmin=fmin,
fmax=fmax,
ref=ref,
amin=amin,
top_db=top_db,
freeze_parameters=True,
)
def preprocess(x):
# normalize raw numpy array then convert it to torch tensor
x = x.astype(float)
x /= np.abs(x).max()
x = x.reshape(1, -1)
x = torch.from_numpy(x).float()
return x