if generated_utterances[g] > num_utterances:
continue
except KeyError:
generated_utterances[g] = 1
if output_html:
def _tqdm(x):
return x
else:
_tqdm = tqdm
# Generate
y_hats = batch_wavegen(model,
c=c,
g=g,
fast=True,
tqdm=_tqdm,
writing_dir=writing_dir)
# Save each utt.
has_ref_file = len(ref_files) > 0
for i, (ref, gen, length) in enumerate(zip(x, y_hats, input_lengths)):
if has_ref_file:
if is_mulaw_quantize(hparams.input_type):
# needs to be float since mulaw_inv returns in range of [-1, 1]
ref = ref.view(-1).float().cpu().numpy()[:length]
else:
ref = ref.view(-1).cpu().numpy()[:length]
gen = gen[:length]
if has_ref_file:
target_audio_path = ref_files[i]
name = splitext(basename(target_audio_path))[0].replace(
if generated_utterances[g] > num_utterances:
continue
except KeyError:
generated_utterances[g] = 1
if output_html:
def _tqdm(x):
return x
else:
_tqdm = tqdm
# Generate
y_hats = batch_wavegen(model,
c=c[:, :, 10:130],
g=g,
fast=True,
tqdm=_tqdm)
# Save each utt.
has_ref_file = len(ref_files) > 0
for i, (ref, gen, length) in enumerate(zip(x, y_hats, input_lengths)):
if has_ref_file:
if is_mulaw_quantize(hparams.input_type):
# needs to be float since mulaw_inv returns in range of [-1, 1]
ref = ref.max(0)[1].view(-1).float().cpu().numpy()[:length]
else:
ref = ref.view(-1).cpu().numpy()[:length]
gen = gen[:length]
if has_ref_file:
target_audio_path = ref_files[i]
try:
generated_utterances[g] += 1
if generated_utterances[g] > num_utterances:
continue
except KeyError:
generated_utterances[g] = 1
if output_html:
def _tqdm(x):
return x
else:
_tqdm = tqdm
# Generate
y_hats = batch_wavegen(model, c=c, g=g, fast=True, tqdm=_tqdm)
# Save each utt.
has_ref_file = len(ref_files) > 0
for i, (ref, gen, length) in enumerate(zip(x, y_hats, input_lengths)):
if has_ref_file:
if is_mulaw_quantize(hparams.input_type):
# needs to be float since mulaw_inv returns in range of [-1, 1]
ref = ref.max(0)[1].view(-1).float().cpu().numpy()[:length]
else:
ref = ref.view(-1).cpu().numpy()[:length]
gen = gen[:length]
if has_ref_file:
target_audio_path = ref_files[i]
name = splitext(basename(target_audio_path))[0].replace(
"-wave", "")
from train import build_model
from synthesis import batch_wavegen
# Model
model = build_model().to(device)
# Load checkpoint
print("Load checkpoint from {}".format(checkpoint_path))
if use_cuda:
checkpoint = torch.load(checkpoint_path)
else:
checkpoint = torch.load(checkpoint_path,
map_location=lambda storage, loc: storage)
model.load_state_dict(checkpoint["state_dict"])
# Generate
y_hat = batch_wavegen(model,
c=None,
g=None,
fast=True,
tqdm=tqdm,
length=length)
gen = y_hat[0, :length]
gen = np.clip(gen, -1.0, 1.0)
# Write, random name
os.makedirs(dst_dir, exist_ok=True)
dst_wav_path = join(dst_dir, "{}_gen.wav".format(random.randint(0, 2**16)))
wavfile.write(dst_wav_path, hparams.sample_rate, to_int16(gen))