def synthesize(speaker, input_file, output_file, params):
print("[Encoding]")
from io_modules.dataset import Dataset
from io_modules.dataset import Encodings
from models.encoder import Encoder
from trainers.encoder import Trainer
encodings = Encodings()
encodings.load('data/models/encoder.encodings')
encoder = Encoder(params, encodings, runtime=True)
encoder.load('data/models/rnn_encoder')
seq = create_lab_input(input_file, speaker)
mgc, att = encoder.generate(seq)
_render_spectrogram(mgc, output_file + '.png')
print("[Vocoding]")
from models.vocoder import Vocoder
from trainers.vocoder import Trainer
vocoder = Vocoder(params, runtime=True)
vocoder.load('data/models/rnn_vocoder')
import time
start = time.time()
signal = vocoder.synthesize(mgc,
batch_size=1000,
temperature=params.temperature)
stop = time.time()
sys.stdout.write(" execution time=" + str(stop - start))
sys.stdout.write('\n')
sys.stdout.flush()
from io_modules.dataset import DatasetIO
dio = DatasetIO()
enc = dio.b16_dec(signal, discreete=True)
dio.write_wave(output_file, enc, params.target_sample_rate)
def load_encoder(params, base_path='data/models'):
from io_modules.dataset import Encodings
from models.encoder import Encoder
encodings = Encodings()
encodings.load('%s/encoder.encodings' % base_path)
encoder = Encoder(params, encodings, runtime=True)
encoder.load('%s/rnn_encoder' % base_path)
return encoder
def phase_3_train_encoder(params):
from io_modules.dataset import Dataset
from io_modules.dataset import Encodings
from models.encoder import Encoder
from trainers.encoder import Trainer
trainset = Dataset("data/processed/train")
devset = Dataset("data/processed/dev")
sys.stdout.write('Found ' + str(len(trainset.files)) +
' training files and ' + str(len(devset.files)) +
' development files\n')
encodings = Encodings()
count = 0
if not params.resume:
for train_file in trainset.files:
count += 1
if count % 100 == 0:
sys.stdout.write('\r' + str(count) + '/' +
str(len(trainset.files)) +
' processed files')
sys.stdout.flush()
from io_modules.dataset import DatasetIO
dio = DatasetIO()
lab_list = dio.read_lab(train_file + ".lab")
for entry in lab_list:
encodings.update(entry)
sys.stdout.write('\r' + str(count) + '/' +
str(len(trainset.files)) + ' processed files\n')
sys.stdout.write('Found ' + str(len(encodings.char2int)) +
' unique symbols, ' +
str(len(encodings.context2int)) +
' unique features and ' +
str(len(encodings.speaker2int)) +
' unique speakers\n')
encodings.store('data/models/encoder.encodings')
else:
encodings.load('data/models/encoder.encodings')
if params.resume:
runtime = True # avoid ortonormal initialization
else:
runtime = False
encoder = Encoder(params, encodings, runtime=runtime)
if params.resume:
sys.stdout.write('Resuming from previous checkpoint\n')
encoder.load('data/models/rnn_encoder')
if params.no_guided_attention:
sys.stdout.write('Disabling guided attention\n')
if params.no_bounds:
sys.stdout.write(
'Using internal stopping condition for synthesis\n')
trainer = Trainer(encoder, trainset, devset)
trainer.start_training(10, 1000, params)
def synthesize(speaker, input_file, output_file, params):
from models.vocoder import device
print(device)
print("[Encoding]")
from io_modules.dataset import Dataset
from io_modules.dataset import Encodings
from models.encoder import Encoder
from trainers.encoder import Trainer
encodings = Encodings()
encodings.load('data/models/encoder.encodings')
encoder = Encoder(params, encodings, runtime=True)
encoder.load('data/models/rnn_encoder')
seq = create_lab_input(input_file, speaker)
mgc, att = encoder.generate(seq)
_render_spectrogram(mgc, output_file + '.png')
print("[Vocoding]")
from models.vocoder import ParallelVocoder
from models.vocoder import Vocoder
vocoder = Vocoder(params)
vocoder.load('data/models/nn_vocoder')
pvocoder = ParallelVocoder(params, vocoder=vocoder)
pvocoder.load('data/models/pnn_vocoder')
import time
start = time.time()
import torch
with torch.no_grad():
signal = pvocoder.synthesize(mgc, batch_size=params.batch_size)
stop = time.time()
sys.stdout.write(" execution time=" + str(stop - start))
sys.stdout.write('\n')
sys.stdout.flush()
from io_modules.dataset import DatasetIO
dio = DatasetIO()
dio.write_wave(output_file, signal / 32768.0, params.target_sample_rate, dtype=signal.dtype)
elif not params.txt_file:
print("Input file is mandatory")
elif not params.output_file:
print("Output file is mandatory")
memory = int(params.memory)
# for compatibility we have to add this paramater
params.learning_rate = 0.0001
dynet_config.set(mem=memory, random_seed=9)
if params.gpu:
dynet_config.set_gpu()
if params.g2p is not None:
from models.g2p import G2P
from io_modules.encodings import Encodings
g2p_encodings = Encodings()
g2p_encodings.load(params.g2p + '.encodings')
g2p = G2P(g2p_encodings)
g2p.load(params.g2p + '-bestAcc.network')
if exists(params.g2p + '.lexicon'):
g2p.load_lexicon(params.g2p + '.lexicon')
else:
g2p = None
synthesize(params.speaker,
params.txt_file,
params.output_file,
params,
g2p=g2p)