def load_dataset(args):
train_dataset = KORDataset(args.data_path, True, 0.1)
test_dataset = KORDataset(args.data_path, False, 0.1)
collate_fn1 = lambda batch: collate_fn_tr(batch, args.max_time_steps, args.
hop_length)
collate_fn2 = lambda batch: collate_fn_synth(batch, args.hop_length)
train_loader = DataLoader(train_dataset,
batch_size=args.bsz,
shuffle=True,
collate_fn=collate_fn1,
num_workers=args.num_workers,
pin_memory=True)
test_loader = DataLoader(test_dataset,
batch_size=args.bsz,
collate_fn=collate_fn1,
num_workers=args.num_workers,
pin_memory=True)
synth_loader = DataLoader(test_dataset,
batch_size=1,
collate_fn=collate_fn2,
num_workers=args.num_workers,
pin_memory=True)
print('num of train samples', len(train_loader))
print('num of test samples', len(test_loader))
return train_loader, test_loader, synth_loader
def load_dataset(args):
collate_fn2 = lambda batch: collate_fn_synth(batch, args.hop_length)
test_dataset = KORDataset(args.data_path, False, 0.1)
synth_loader = DataLoader(test_dataset, batch_size=1, collate_fn=collate_fn2,
num_workers=args.num_workers, pin_memory=True)
print('sr', args.sr)
return synth_loader
def load_dataset(args):
train_dataset = KORDataset(args.data_path, True, 0.1)
test_dataset = KORDataset(args.data_path, False, 0.1)
if args.distributed:
train_sampler = torch.utils.data.distributed.DistributedSampler(train_dataset)
else:
train_sampler = None
collate_fn1 = lambda batch: collate_fn_tr(batch, args.max_time_steps, args.hop_length)
collate_fn2 = lambda batch: collate_fn_synth(batch, args.hop_length)
train_loader = DataLoader(train_dataset, batch_size=args.bsz, shuffle=(train_sampler is None), collate_fn=collate_fn1,
num_workers=args.num_workers, pin_memory=True, sampler=train_sampler)
test_loader = DataLoader(test_dataset, batch_size=args.bsz, collate_fn=collate_fn1,
num_workers=args.num_workers, pin_memory=True)
synth_loader = DataLoader(test_dataset, batch_size=1, collate_fn=collate_fn2,
num_workers=args.num_workers, pin_memory=True)
print('num of train samples', len(train_loader))
print('num of test samples', len(test_loader))
return train_loader, test_loader, synth_loader
def load_dataset(args):
train_dataset = KORDataset(args.data_path, True, 0.1)
test_dataset = KORDataset(args.data_path, False, 0.1)
train_loader = DataLoader(train_dataset,
batch_size=args.batch_size,
shuffle=True,
collate_fn=collate_fn,
num_workers=args.num_workers,
pin_memory=True)
test_loader = DataLoader(test_dataset,
batch_size=args.batch_size,
collate_fn=collate_fn,
num_workers=args.num_workers,
pin_memory=True)
synth_loader = DataLoader(test_dataset,
batch_size=1,
collate_fn=collate_fn_synthesize,
num_workers=args.num_workers,
pin_memory=True)
print('num of train samples', len(train_loader))
print('num of test samples', len(test_loader))
return train_loader, test_loader, synth_loader
# Checkpoint dir
if not os.path.isdir(args.save):
os.makedirs(args.save)
if not os.path.isdir(args.loss):
os.makedirs(args.loss)
if not os.path.isdir(args.sample_path):
os.makedirs(args.sample_path)
if not os.path.isdir(os.path.join(args.save, args.model_name)):
os.makedirs(os.path.join(args.save, args.model_name))
use_cuda = torch.cuda.is_available()
device = torch.device("cuda" if use_cuda else "cpu")
# LOAD DATASETS
train_dataset = KORDataset(args.data_path, True, 0.1)
test_dataset = KORDataset(args.data_path, False, 0.1)
# collate_fn1 = lambda batch: collate_fn_tr(batch, 16000, 256)
train_loader = DataLoader(train_dataset,
batch_size=args.batch_size,
shuffle=True,
collate_fn=collate_fn,
num_workers=args.num_workers,
pin_memory=True)
test_loader = DataLoader(test_dataset,
batch_size=args.batch_size,
collate_fn=collate_fn,
num_workers=args.num_workers,
pin_memory=True)