def __getitem__(self, index):
path = self.file_path + '/' + self.file_names[index] + '.wav'
data = librosa.core.load(path, sr=SR, mono=True)[0]
#pre-emphasis
data = preprocessing.preemphasis(data)
mfcc = librosa.feature.mfcc(y=data, sr=SR, hop_length=1024, n_mfcc=40)
if self.transform:
mfcc = self.transform(mfcc)
soundFormatted = mfcc.float()
soundFormatted = torch.cat((soundFormatted, soundFormatted,
soundFormatted)) #?? size 확인하기
# print("Brinda", soundFormatted.shape)
return soundFormatted, self.labels[index]
#print("soundformat shape:", soundFormatted.shape)
soundFormatted = torch.from_numpy(mfcc).float()
soundFormatted = torch.cat(
(soundFormatted, soundFormatted, soundFormatted)) #?? size 확인하기
# soundFormatted = torch.unsqueeze(soundFormatted, dim=0)
# soundFormatted = torch.cat((soundFormatted,soundFormatted,soundFormatted))
return soundFormatted, self.labels[index]
def __getitem__(self, index):
path = self.file_path + '/' + self.file_names[index] + '.wav'
data = librosa.core.load(path, sr=SR, mono=True)[0]
#pre-emphasis
data = preprocessing.preemphasis(data)
mrcg = MRCG.mrcg_extract(data, SR)
# if self.transform:
# mrcg = self.transform(mrcg)
# soundFormatted = mrcg.float()
# # soundFormatted = torch.squeeze(soundFormatted)
# # soundFormatted.transpose_(0, 1)
# print("mrcg shape", soundFormatted.shape)
# return soundFormatted, self.labels[index]
soundFormatted = torch.from_numpy(mrcg).float()
# soundFormatted.transpose_(0, 1)
soundFormatted = soundFormatted.unsqueeze(dim=0)
soundFormatted = torch.cat(
(soundFormatted, soundFormatted, soundFormatted))
# print("mrgc shape", soundFormatted.shape)
return soundFormatted, self.labels[index]
def __getitem__(self, index):
path = self.file_path + '/' + self.file_names[index] + '.wav'
data = librosa.core.load(path, sr=SR, mono=True)[0]
#pre-emphasis
data = preprocessing.preemphasis(data)
mfcc = librosa.feature.mfcc(y=data, sr=SR, hop_length=1024, n_mfcc=40)
soundFormatted = torch.from_numpy(mfcc).float()
return soundFormatted, self.file_names[index]
def __getitem__(self, index):
path = self.file_path + '/' + self.file_names[index] + '.wav'
data = librosa.core.load(path, sr=SR, mono=True)[0]
#pre-emphasis
data = preprocessing.preemphasis(data)
mrcg = MRCG.mrcg_extract(data, SR)
soundFormatted = torch.from_numpy(mrcg).float()
# soundFormatted.transpose_(0, 1)
# print("data shape", soundFormatted.shape)
return soundFormatted, self.file_names[index]
def __getitem__(self, index):
path = self.file_path + '/' + self.file_names[index] + '.wav'
data = librosa.core.load(path, sr=SR, mono=True)[0]
#pre-emphasis
data = preprocessing.preemphasis(data)
mfcc = librosa.feature.mfcc(y=data, sr=SR, hop_length=1024, n_mfcc=40)
if self.transform:
mfcc = self.transform(mfcc)
soundFormatted = mfcc.float()
soundFormatted = torch.squeeze(soundFormatted)
# soundFormatted.transpose_(0, 1)
# print("data shape", soundFormatted.shape)
return soundFormatted, self.labels[index]
soundFormatted = torch.from_numpy(mfcc).float()
# soundFormatted.transpose_(0, 1)
# print("data shape", soundFormatted.shape)
return soundFormatted, self.labels[index]