# Dataset and Dataloader setup
dataset = PyTorchDataset(X, Mel, Y)
data_loader = data_utils.DataLoader(dataset,
batch_size=hparams.batch_size,
num_workers=hparams.num_workers,
shuffle=True,
collate_fn=collate_fn,
pin_memory=hparams.pin_memory)
# Model
model = Tacotron(
n_vocab=1 + len(charids),
embedding_dim=256,
mel_dim=hparams.num_mels,
linear_dim=hparams.num_freq,
r=hparams.outputs_per_step,
padding_idx=hparams.padding_idx,
use_memory_mask=hparams.use_memory_mask,
)
model = model.cuda()
optimizer = optim.Adam(model.parameters(),
lr=hparams.initial_learning_rate,
betas=(hparams.adam_beta1, hparams.adam_beta2),
weight_decay=hparams.weight_decay)
# Load checkpoint
if checkpoint_path:
print("Load checkpoint from: {}".format(checkpoint_path))
checkpoint = torch.load(checkpoint_path)
checkpoint_path = args["<checkpoint>"]
text_list_file_path = args["<text_list_file>"]
dst_dir = args["<dst_dir>"]
max_decoder_steps = int(args["--max-decoder-steps"])
file_name_suffix = args["--file-name-suffix"]
checkpoint = torch.load(checkpoint_path)
checkpoints_dir = os.path.dirname(checkpoint_path)
with open(checkpoints_dir + '/ids.json') as f:
charids = json.load(f)
model = Tacotron(
n_vocab=len(charids) + 1,
embedding_dim=256,
mel_dim=hparams.num_mels,
linear_dim=hparams.num_freq,
r=hparams.outputs_per_step,
padding_idx=hparams.padding_idx,
use_memory_mask=hparams.use_memory_mask,
)
checkpoint = torch.load(checkpoint_path)
checkpoints_dir = os.path.dirname(checkpoint_path)
with open(checkpoints_dir + '/ids.json') as f:
charids = json.load(f)
charids = dict(charids)
model.load_state_dict(checkpoint["state_dict"])
model.decoder.max_decoder_steps = max_decoder_steps
os.makedirs(dst_dir, exist_ok=True)