def setup(self):
# load configs
self.TTS_CONFIG = load_config(self.TTS_CONFIG)
self.VOCODER_CONFIG = load_config(self.VOCODER_CONFIG)
# load the audio processor
self.ap = AudioProcessor(**self.TTS_CONFIG.audio)
# load the model
num_chars = len(phonemes) if self.TTS_CONFIG.use_phonemes else len(
symbols)
self.model = setup_model(num_chars, len(self.speakers),
self.TTS_CONFIG)
self.model, _ = load_checkpoint(self.model,
self.TTS_MODEL,
use_cuda=self.use_cuda)
self.model.eval()
# LOAD VOCODER MODEL
self.vocoder_model = setup_generator(self.VOCODER_CONFIG)
self.vocoder_model, _ = load_vocoder_checkpoint(
self.vocoder_model, checkpoint_path=self.VOCODER_MODEL)
self.vocoder_model.remove_weight_norm()
self.vocoder_model.inference_padding = 0
self.ap_vocoder = AudioProcessor(**self.VOCODER_CONFIG['audio'])
if self.use_cuda:
self.vocoder_model.cuda()
self.vocoder_model.eval()
def load_vocoder(self, model_file, model_config, use_cuda):
self.vocoder_config = load_config(model_config)
self.vocoder_model = setup_generator(self.vocoder_config)
self.vocoder_model.load_state_dict(
torch.load(model_file, map_location="cpu")["model"])
self.vocoder_model.remove_weight_norm()
self.vocoder_model.inference_padding = 0
self.vocoder_config = load_config(model_config)
if use_cuda:
self.vocoder_model.cuda()
self.vocoder_model.eval()
def details():
model_config = load_config(args.tts_config)
if args.vocoder_config is not None and os.path.isfile(args.vocoder_config):
vocoder_config = load_config(args.vocoder_config)
else:
vocoder_config = None
return render_template('details.html',
show_details=args.show_details,
model_config=model_config,
vocoder_config=vocoder_config,
args=args.__dict__)
def setup():
use_cuda = True
# model paths
TTS_MODEL = "tts_model.pth.tar"
TTS_CONFIG = "config.json"
VOCODER_MODEL = "vocoder_model.pth.tar"
VOCODER_CONFIG = "config_vocoder.json"
# Load configs
TTS_CONFIG = load_config(TTS_CONFIG)
VOCODER_CONFIG = load_config(VOCODER_CONFIG)
ap = AudioProcessor(**TTS_CONFIG.audio)
# LOAD TTS MODEL
# multi speaker
speaker_id = None
speakers = []
# load the model
num_chars = len(phonemes) if TTS_CONFIG.use_phonemes else len(symbols)
model = setup_model(num_chars, len(speakers), TTS_CONFIG)
# load model state
cp = torch.load(TTS_MODEL, map_location=torch.device('cpu'))
# load the model
model.load_state_dict(cp['model'])
if use_cuda:
model.cuda()
model.eval()
# set model stepsize
if 'r' in cp:
model.decoder.set_r(cp['r'])
from TTS.vocoder.utils.generic_utils import setup_generator
# LOAD VOCODER MODEL
vocoder_model = setup_generator(VOCODER_CONFIG)
vocoder_model.load_state_dict(torch.load(VOCODER_MODEL, map_location="cpu")["model"])
vocoder_model.remove_weight_norm()
vocoder_model.inference_padding = 0
ap_vocoder = AudioProcessor(**VOCODER_CONFIG['audio'])
if use_cuda:
vocoder_model.cuda()
vocoder_model.eval()
return model, vocoder_model, speaker_id, TTS_CONFIG, use_cuda, ap
def do_phonemize(args):
"""Generate phonemes for text using config"""
from TTS.utils.io import load_config
from TTS.tts.utils.text import make_symbols, phoneme_to_sequence
c = load_config(args.config)
_, phonemes = make_symbols(**c.characters)
if args.text:
# Use arguments
texts = args.text
else:
# Use stdin
texts = sys.stdin
if os.isatty(sys.stdin.fileno()):
print("Reading text from stdin...", file=sys.stderr)
for line in texts:
line = line.strip()
if not line:
continue
line_indexes = phoneme_to_sequence(
line,
[c.text_cleaner],
language=c.phoneme_language,
enable_eos_bos=False,
tp=c.characters if "characters" in c.keys() else None,
backend=c.phoneme_backend,
)
line_phonemes = [phonemes[i] for i in line_indexes]
print(args.separator.join(line_phonemes))
def load_vocoder(lib_path, model_file, model_config, use_cuda):
sys.path.append(lib_path) # set this if ParallelWaveGAN is not installed globally
#pylint: disable=import-outside-toplevel
vocoder_config = load_config(model_config)
vocoder_model = setup_generator(vocoder_config)
checkpoint = torch.load(model_file, map_location='cpu')
print(' > Model step:', checkpoint['step'])
vocoder_model.load_state_dict(checkpoint['model'])
vocoder_model.remove_weight_norm()
vocoder_model.inference_padding = 0
vocoder_config = load_config(model_config)
ap_vocoder = AudioProcessor(**vocoder_config['audio'])
if use_cuda:
vocoder_model.cuda()
return vocoder_model.eval(), ap_vocoder
def load_wavernn(self, lib_path, model_file, model_config, use_cuda):
# TODO: set a function in wavernn code base for model setup and call it here.
sys.path.append(
lib_path) # set this if WaveRNN is not installed globally
#pylint: disable=import-outside-toplevel
from WaveRNN.models.wavernn import Model
print(" > Loading WaveRNN model ...")
print(" | > model config: ", model_config)
print(" | > model file: ", model_file)
self.wavernn_config = load_config(model_config)
# This is the default architecture we use for our models.
# You might need to update it
self.wavernn = Model(
rnn_dims=512,
fc_dims=512,
mode=self.wavernn_config.mode,
mulaw=self.wavernn_config.mulaw,
pad=self.wavernn_config.pad,
use_aux_net=self.wavernn_config.use_aux_net,
use_upsample_net=self.wavernn_config.use_upsample_net,
upsample_factors=self.wavernn_config.upsample_factors,
feat_dims=80,
compute_dims=128,
res_out_dims=128,
res_blocks=10,
hop_length=self.ap.hop_length,
sample_rate=self.ap.sample_rate,
).cuda()
check = torch.load(model_file, map_location="cpu")
self.wavernn.load_state_dict(check['model'])
if use_cuda:
self.wavernn.cuda()
self.wavernn.eval()
def load_tacotron2(use_cuda):
"""
Loads the Tacotron2 model
Parameters
----------
use_cuda : bool
whether to use the gpu
Returns
-------
model, audio processor, model config
"""
TTS_MODEL = model_path / 'model.pth.tar'
TTS_CONFIG = model_path / 'config.json'
TTS_CONFIG = load_config(TTS_CONFIG)
TTS_CONFIG.audio['stats_path'] = str(model_path / 'scale_stats.npy')
ap = AudioProcessor(**TTS_CONFIG.audio)
num_chars = len(phonemes) if TTS_CONFIG.use_phonemes else len(symbols)
model = setup_model(num_chars, 0, TTS_CONFIG)
cp = torch.load(TTS_MODEL, map_location=torch.device('cpu'))
model.load_state_dict(cp['model'])
if use_cuda:
model.cuda()
if 'r' in cp:
model.decoder.set_r(cp['r'])
model.eval()
return model, ap, TTS_CONFIG
def load_vocoder(use_cuda):
"""
Loads the Vocoder model
Parameters
----------
use_cuda : bool
whether to use the gpu
Returns
-------
model, audio processor, model config
"""
VOCODER_MODEL = model_path / 'vocoder_model.pth.tar'
VOCODER_CONFIG = model_path / 'vocoder_config.json'
VOCODER_CONFIG = load_config(VOCODER_CONFIG)
VOCODER_CONFIG.audio['stats_path'] = str(model_path /
'vocoder_scale_stats.npy')
ap_vocoder = AudioProcessor(**VOCODER_CONFIG['audio'])
vocoder_model = setup_generator(VOCODER_CONFIG)
cp = torch.load(VOCODER_MODEL, map_location=torch.device('cpu'))
vocoder_model.load_state_dict(cp['model'])
vocoder_model.remove_weight_norm()
vocoder_model.inference_padding = 0
if use_cuda:
vocoder_model.cuda()
vocoder_model.eval()
return vocoder_model, ap_vocoder, VOCODER_CONFIG
def load_vocoder(self, model_file, model_config, use_cuda):
self.vocoder_config = load_config(model_config)
self.vocoder_ap = AudioProcessor(**self.vocoder_config['audio'])
self.vocoder_model = setup_generator(self.vocoder_config)
self.vocoder_model.load_checkpoint(self.vocoder_config, model_file, eval=True)
if use_cuda:
self.vocoder_model.cuda()
def test_in_out(self):
self._create_random_model()
config = load_config(os.path.join(get_tests_input_path(), 'server_config.json'))
tts_root_path = get_tests_output_path()
config['tts_checkpoint'] = os.path.join(tts_root_path, config['tts_checkpoint'])
config['tts_config'] = os.path.join(tts_root_path, config['tts_config'])
synthesizer = Synthesizer(config['tts_checkpoint'], config['tts_config'], None, None)
synthesizer.tts("Better this test works!!")
def init_speaker_encoder(self, model_path: str, config_path: str) -> None:
self.speaker_encoder_config = load_config(config_path)
self.speaker_encoder = setup_model(self.speaker_encoder_config)
self.speaker_encoder.load_checkpoint(config_path, model_path, True)
self.speaker_encoder_ap = AudioProcessor(
**self.speaker_encoder_config.audio)
# normalize the input audio level and trim silences
self.speaker_encoder_ap.do_sound_norm = True
self.speaker_encoder_ap.do_trim_silence = True
def download_model(self, model_name):
"""Download model files given the full model name.
Model name is in the format
'type/language/dataset/model'
e.g. 'tts_model/en/ljspeech/tacotron'
Every model must have the following files
- *.pth.tar : pytorch model checkpoint file.
- config.json : model config file.
- scale_stats.npy (if exist): scale values for preprocessing.
Args:
model_name (str): model name as explained above.
TODO: support multi-speaker models
"""
# fetch model info from the dict
model_type, lang, dataset, model = model_name.split("/")
model_full_name = f"{model_type}--{lang}--{dataset}--{model}"
model_item = self.models_dict[model_type][lang][dataset][model]
# set the model specific output path
output_path = os.path.join(self.output_prefix, model_full_name)
output_model_path = os.path.join(output_path, "model_file.pth.tar")
output_config_path = os.path.join(output_path, "config.json")
if os.path.exists(output_path):
print(f" > {model_name} is already downloaded.")
else:
os.makedirs(output_path, exist_ok=True)
print(f" > Downloading model to {output_path}")
output_stats_path = os.path.join(output_path, 'scale_stats.npy')
# download files to the output path
if self._check_dict_key(model_item, 'github_rls_url'):
# download from github release
# TODO: pass output_path
self._download_zip_file(model_item['github_rls_url'],
output_path)
else:
# download from gdrive
self._download_gdrive_file(model_item['model_file'],
output_model_path)
self._download_gdrive_file(model_item['config_file'],
output_config_path)
if self._check_dict_key(model_item, 'stats_file'):
self._download_gdrive_file(model_item['stats_file'],
output_stats_path)
# set the scale_path.npy file path in the model config.json
if self._check_dict_key(
model_item,
'stats_file') or os.path.exists(output_stats_path):
# set scale stats path in config.json
config_path = output_config_path
config = load_config(config_path)
config["audio"]['stats_path'] = output_stats_path
with open(config_path, "w") as jf:
json.dump(config, jf)
return output_model_path, output_config_path, model_item
def __init__(self, use_cuda=False, verbose=False):
self.use_cuda = use_cuda
self.verbose = verbose
# load configs
self.TTS_CONFIG = load_config(TTS_CONFIG)
self.VOCODER_CONFIG = load_config(VOCODER_CONFIG)
# load the audio processor
self.ap = AudioProcessor(**self.TTS_CONFIG.audio)
# LOAD TTS MODEL
self.speaker_id = None
speakers = []
# load the model
num_chars = len(phonemes) if self.TTS_CONFIG.use_phonemes else len(symbols)
self.model = setup_model(num_chars, len(speakers), self.TTS_CONFIG)
# load model state
cp = torch.load(TTS_MODEL, map_location=torch.device('cpu'))
# load the model
self.model.load_state_dict(cp['model'])
if self.use_cuda:
self.model.cuda()
self.model.eval()
# set model stepsize
if 'r' in cp:
self.model.decoder.set_r(cp['r'])
# LOAD VOCODER MODEL
self.vocoder_model = setup_generator(self.VOCODER_CONFIG)
self.vocoder_model.load_state_dict(torch.load(VOCODER_MODEL, map_location="cpu")["model"])
self.vocoder_model.remove_weight_norm()
self.vocoder_model.inference_padding = 0
ap_vocoder = AudioProcessor(**self.VOCODER_CONFIG['audio'])
if self.use_cuda:
self.vocoder_model.cuda()
self.vocoder_model.eval()
def __init__(self, TTS_MODEL, TTS_CONFIG, VOCODER_MODEL, VOCODER_CONFIG,
use_cuda, use_gl):
self.use_cuda = use_cuda
self.use_gl = use_gl
# model paths
self.tts_config = load_config(TTS_CONFIG)
vocoder_config = load_config(VOCODER_CONFIG)
# load audio processor
self.ap = AudioProcessor(**self.tts_config.audio)
# LOAD TTS MODEL
# multi speaker
self.speaker_id = None
speakers = []
# load the model
num_chars = len(phonemes) if self.tts_config.use_phonemes else len(
symbols)
self.model = setup_model(num_chars, len(speakers), self.tts_config)
# load model state
self.cp = torch.load(TTS_MODEL, map_location=torch.device('cpu'))
# load the model
self.model.load_state_dict(self.cp['model'])
if self.use_cuda:
self.model.cuda()
self.model.train(False)
self.model.eval()
# set model stepsize
if 'r' in self.cp:
self.model.decoder.set_r(self.cp['r'])
# LOAD VOCODER MODEL
self.vocoder_model = setup_generator(vocoder_config)
self.vocoder_model.load_state_dict(
torch.load(VOCODER_MODEL, map_location="cpu")["model"])
self.vocoder_model.remove_weight_norm()
self.vocoder_model.inference_padding = 0
#ap_vocoder = AudioProcessor(**vocoder_config['audio'])
if use_cuda:
self.vocoder_model.cuda()
self.vocoder_model.train(False)
self.vocoder_model.eval()
#get sample rate
self.sample_rate = self.ap.sample_rate
gc.collect(2)
def _create_random_model(self):
# pylint: disable=global-statement
global symbols, phonemes
config = load_config(os.path.join(get_tests_output_path(), 'dummy_model_config.json'))
if 'characters' in config.keys():
symbols, phonemes = make_symbols(**config.characters)
num_chars = len(phonemes) if config.use_phonemes else len(symbols)
model = setup_model(num_chars, 0, config)
output_path = os.path.join(get_tests_output_path())
save_checkpoint(model, None, 10, 10, 1, output_path)
def _load_vocoder(self, model_file: str, model_config: str,
use_cuda: bool) -> None:
"""Load the vocoder model.
Args:
model_file (str): path to the model checkpoint.
model_config (str): path to the model config file.
use_cuda (bool): enable/disable CUDA use.
"""
self.vocoder_config = load_config(model_config)
self.vocoder_ap = AudioProcessor(verbose=False,
**self.vocoder_config["audio"])
self.vocoder_model = setup_generator(self.vocoder_config)
self.vocoder_model.load_checkpoint(self.vocoder_config,
model_file,
eval=True)
if use_cuda:
self.vocoder_model.cuda()
def _load_tts(self, tts_checkpoint: str, tts_config_path: str,
use_cuda: bool) -> None:
"""Load the TTS model.
Args:
tts_checkpoint (str): path to the model checkpoint.
tts_config_path (str): path to the model config file.
use_cuda (bool): enable/disable CUDA use.
"""
# pylint: disable=global-statement
global symbols, phonemes
self.tts_config = load_config(tts_config_path)
self.use_phonemes = self.tts_config.use_phonemes
self.ap = AudioProcessor(verbose=False, **self.tts_config.audio)
if "characters" in self.tts_config.keys():
symbols, phonemes = make_symbols(**self.tts_config.characters)
if self.use_phonemes:
self.input_size = len(phonemes)
else:
self.input_size = len(symbols)
if self.tts_config.use_speaker_embedding is True:
self.tts_speakers_file = (
self.tts_speakers_file if self.tts_speakers_file else
self.tts_config["external_speaker_embedding_file"])
self._load_speakers(self.tts_speakers_file)
self.tts_model = setup_model(
self.input_size,
num_speakers=self.num_speakers,
c=self.tts_config,
speaker_embedding_dim=self.speaker_embedding_dim,
)
self.tts_model.load_checkpoint(self.tts_config,
tts_checkpoint,
eval=True)
if use_cuda:
self.tts_model.cuda()
def load_tts(self, tts_checkpoint, tts_config, use_cuda):
# pylint: disable=global-statement
global symbols, phonemes
self.tts_config = load_config(tts_config)
self.use_phonemes = self.tts_config.use_phonemes
self.ap = AudioProcessor(**self.tts_config.audio)
if 'characters' in self.tts_config.keys():
symbols, phonemes = make_symbols(**self.tts_config.characters)
if self.use_phonemes:
self.input_size = len(phonemes)
else:
self.input_size = len(symbols)
self.tts_model = setup_model(self.input_size, num_speakers=self.num_speakers, c=self.tts_config)
self.tts_model.load_checkpoint(tts_config, tts_checkpoint, eval=True)
if use_cuda:
self.tts_model.cuda()
def download_model(self, model_name):
"""Download model files given the full model name.
Model name is in the format
'type/language/dataset/model'
e.g. 'tts_model/en/ljspeech/tacotron'
Args:
model_name (str): model name as explained above.
TODO: support multi-speaker models
"""
# fetch model info from the dict
model_type, lang, dataset, model = model_name.split("/")
model_full_name = f"{model_type}--{lang}--{dataset}--{model}"
model_item = self.models_dict[model_type][lang][dataset][model]
# set the model specific output path
output_path = os.path.join(self.output_prefix, model_full_name)
output_model_path = os.path.join(output_path, "model_file.pth.tar")
output_config_path = os.path.join(output_path, "config.json")
if os.path.exists(output_path):
print(f" > {model_name} is already downloaded.")
else:
os.makedirs(output_path, exist_ok=True)
print(f" > Downloading model to {output_path}")
output_stats_path = None
# download files to the output path
self._download_file(model_item['model_file'], output_model_path)
self._download_file(model_item['config_file'], output_config_path)
if model_item['stats_file'] is not None and len(
model_item['stats_file']) > 1:
output_stats_path = os.path.join(output_path,
'scale_stats.npy')
self._download_file(model_item['stats_file'],
output_stats_path)
# set scale stats path in config.json
config_path = output_config_path
config = load_config(config_path)
config["audio"]['stats_path'] = output_stats_path
with open(config_path, "w") as jf:
json.dump(config, jf)
return output_model_path, output_config_path
def load_tts(self, tts_checkpoint, tts_config, use_cuda):
# pylint: disable=global-statement
global symbols, phonemes
print(" > Loading TTS model ...")
print(" | > model config: ", tts_config)
print(" | > checkpoint file: ", tts_checkpoint)
self.tts_config = load_config(tts_config)
self.use_phonemes = self.tts_config.use_phonemes
self.ap = AudioProcessor(**self.tts_config.audio)
if 'characters' in self.tts_config.keys():
symbols, phonemes = make_symbols(**self.tts_config.characters)
if self.use_phonemes:
self.input_size = len(phonemes)
else:
self.input_size = len(symbols)
# TODO: fix this for multi-speaker model - load speakers
if self.config.tts_speakers is not None:
self.tts_speakers = load_speaker_mapping(self.config.tts_speakers)
num_speakers = len(self.tts_speakers)
else:
num_speakers = 0
self.tts_model = setup_model(self.input_size,
num_speakers=num_speakers,
c=self.tts_config)
# load model state
cp = torch.load(tts_checkpoint, map_location=torch.device('cpu'))
# load the model
self.tts_model.load_state_dict(cp['model'])
if use_cuda:
self.tts_model.cuda()
self.tts_model.eval()
self.tts_model.decoder.max_decoder_steps = 3000
if 'r' in cp:
self.tts_model.decoder.set_r(cp['r'])
print(f" > model reduction factor: {cp['r']}")
def test_speaker_embedding():
# load config
config = load_config(encoder_config_path)
config["audio"]["resample"] = True
# create a dummy speaker encoder
model = SpeakerEncoder(**config.model)
save_checkpoint(model, None, None, get_tests_input_path(), 0, 0)
# load audio processor and speaker encoder
ap = AudioProcessor(**config.audio)
manager = SpeakerManager(encoder_model_path=encoder_model_path,
encoder_config_path=encoder_config_path)
# load a sample audio and compute embedding
waveform = ap.load_wav(sample_wav_path)
mel = ap.melspectrogram(waveform)
x_vector = manager.compute_x_vector(mel.T)
assert x_vector.shape[1] == 256
# compute x_vector directly from an input file
x_vector = manager.compute_x_vector_from_clip(sample_wav_path)
x_vector2 = manager.compute_x_vector_from_clip(sample_wav_path)
x_vector = torch.FloatTensor(x_vector)
x_vector2 = torch.FloatTensor(x_vector2)
assert x_vector.shape[0] == 256
assert (x_vector - x_vector2).sum() == 0.0
# compute x_vector from a list of wav files.
x_vector3 = manager.compute_x_vector_from_clip(
[sample_wav_path, sample_wav_path2])
x_vector3 = torch.FloatTensor(x_vector3)
assert x_vector3.shape[0] == 256
assert (x_vector - x_vector3).sum() != 0.0
# remove dummy model
os.remove(encoder_model_path)
import os
import unittest
import torch as T
from tests import get_tests_input_path
from TTS.speaker_encoder.losses import GE2ELoss, AngleProtoLoss
from TTS.speaker_encoder.model import SpeakerEncoder
from TTS.utils.io import load_config
file_path = get_tests_input_path()
c = load_config(os.path.join(file_path, "test_config.json"))
class SpeakerEncoderTests(unittest.TestCase):
# pylint: disable=R0201
def test_in_out(self):
dummy_input = T.rand(4, 20, 80) # B x T x D
dummy_hidden = [T.rand(2, 4, 128), T.rand(2, 4, 128)]
model = SpeakerEncoder(input_dim=80,
proj_dim=256,
lstm_dim=768,
num_lstm_layers=3)
# computing d vectors
output = model.forward(dummy_input)
assert output.shape[0] == 4
assert output.shape[1] == 256
output = model.inference(dummy_input)
assert output.shape[0] == 4
assert output.shape[1] == 256
# compute d vectors by passing LSTM hidden
import argparse
from TTS.utils.io import load_config
from TTS.vocoder.tf.utils.generic_utils import setup_generator
from TTS.vocoder.tf.utils.io import load_checkpoint
from TTS.vocoder.tf.utils.tflite import convert_melgan_to_tflite
parser = argparse.ArgumentParser()
parser.add_argument("--tf_model",
type=str,
help="Path to target torch model to be converted to TF.")
parser.add_argument("--config_path",
type=str,
help="Path to config file of torch model.")
parser.add_argument("--output_path",
type=str,
help="path to tflite output binary.")
args = parser.parse_args()
# Set constants
CONFIG = load_config(args.config_path)
# load the model
model = setup_generator(CONFIG)
model.build_inference()
model = load_checkpoint(model, args.tf_model)
# create tflite model
tflite_model = convert_melgan_to_tflite(model, output_path=args.output_path)
default="",
help='DISTRIBUTED: process group id.')
args = parser.parse_args()
if args.continue_path != '':
args.output_path = args.continue_path
args.config_path = os.path.join(args.continue_path, 'config.json')
list_of_files = glob.glob(
args.continue_path +
"/*.pth.tar") # * means all if need specific format then *.csv
latest_model_file = max(list_of_files, key=os.path.getctime)
args.restore_path = latest_model_file
print(f" > Training continues for {args.restore_path}")
# setup output paths and read configs
c = load_config(args.config_path)
# check_config(c)
_ = os.path.dirname(os.path.realpath(__file__))
# DISTRIBUTED
if c.mixed_precision:
print(" > Mixed precision is enabled")
OUT_PATH = args.continue_path
if args.continue_path == '':
OUT_PATH = create_experiment_folder(c.output_path, c.run_name,
args.debug)
AUDIO_PATH = os.path.join(OUT_PATH, 'test_audios')
c_logger = ConsoleLogger()
"if CONFIG.use_external_speaker_embedding_file is true, name of speaker embedding reference file present in speakers.json, else target speaker_fileid if the model is multi-speaker.",
default=None)
parser.add_argument('--gst_style',
help="Wav path file for GST stylereference.",
default=None)
parser.add_argument(
'--save_spectogram',
type=bool,
help=
"If true save raw spectogram for further (vocoder) processing in out_path.",
default=False)
args = parser.parse_args()
# load the config
C = load_config(args.config_path)
C.forward_attn_mask = True
# load the audio processor
ap = AudioProcessor(**C.audio)
# if the vocabulary was passed, replace the default
if 'characters' in C.keys():
symbols, phonemes = make_symbols(**C.characters)
speaker_embedding = None
speaker_embedding_dim = None
num_speakers = 0
# load speakers
if args.speakers_json != '':
import torch
from torch import nn, optim
from tests import get_tests_input_path
from TTS.tts.layers.losses import MSELossMasked
from TTS.tts.models.tacotron2 import Tacotron2
from TTS.utils.audio import AudioProcessor
from TTS.utils.io import load_config
# pylint: disable=unused-variable
torch.manual_seed(1)
use_cuda = torch.cuda.is_available()
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
c = load_config(os.path.join(get_tests_input_path(), "test_config.json"))
ap = AudioProcessor(**c.audio)
WAV_FILE = os.path.join(get_tests_input_path(), "example_1.wav")
class TacotronTrainTest(unittest.TestCase):
def test_train_step(self): # pylint: disable=no-self-use
input_dummy = torch.randint(0, 24, (8, 128)).long().to(device)
input_lengths = torch.randint(100, 128, (8, )).long().to(device)
input_lengths = torch.sort(input_lengths, descending=True)[0]
mel_spec = torch.rand(8, 30, c.audio["num_mels"]).to(device)
mel_postnet_spec = torch.rand(8, 30, c.audio["num_mels"]).to(device)
mel_lengths = torch.randint(20, 30, (8, )).long().to(device)
mel_lengths[0] = 30
stop_targets = torch.zeros(8, 30, 1).float().to(device)
def process_args(args, model_class):
"""Process parsed comand line arguments based on model class (tts or vocoder).
Args:
args (argparse.Namespace or dict like): Parsed input arguments.
model_type (str): Model type used to check config parameters and setup
the TensorBoard logger. One of ['tts', 'vocoder'].
Raises:
ValueError: If `model_type` is not one of implemented choices.
Returns:
c (TTS.utils.io.AttrDict): Config paramaters.
out_path (str): Path to save models and logging.
audio_path (str): Path to save generated test audios.
c_logger (TTS.utils.console_logger.ConsoleLogger): Class that does
logging to the console.
tb_logger (TTS.utils.tensorboard.TensorboardLogger): Class that does
the TensorBoard loggind.
"""
if args.continue_path:
args.output_path = args.continue_path
args.config_path = os.path.join(args.continue_path, "config.json")
args.restore_path, best_model = get_last_checkpoint(args.continue_path)
if not args.best_path:
args.best_path = best_model
# setup output paths and read configs
c = load_config(args.config_path)
_ = os.path.dirname(os.path.realpath(__file__))
if 'mixed_precision' in c and c.mixed_precision:
print(" > Mixed precision mode is ON")
out_path = args.continue_path
if not out_path:
out_path = create_experiment_folder(c.output_path, c.run_name,
args.debug)
audio_path = os.path.join(out_path, "test_audios")
c_logger = ConsoleLogger()
tb_logger = None
if args.rank == 0:
os.makedirs(audio_path, exist_ok=True)
new_fields = {}
if args.restore_path:
new_fields["restore_path"] = args.restore_path
new_fields["github_branch"] = get_git_branch()
# if model characters are not set in the config file
# save the default set to the config file for future
# compatibility.
if model_class == 'tts' and 'characters' not in c:
used_characters = parse_symbols()
new_fields['characters'] = used_characters
copy_model_files(c, args.config_path, out_path, new_fields)
os.chmod(audio_path, 0o775)
os.chmod(out_path, 0o775)
log_path = out_path
tb_logger = TensorboardLogger(log_path, model_name=model_class.upper())
# write model desc to tensorboard
tb_logger.tb_add_text("model-description", c["run_description"], 0)
return c, out_path, audio_path, c_logger, tb_logger
parser = argparse.ArgumentParser()
parser.add_argument('--torch_model_path',
type=str,
help='Path to target torch model to be converted to TF.')
parser.add_argument('--config_path',
type=str,
help='Path to config file of torch model.')
parser.add_argument(
'--output_path',
type=str,
help='path to output file including file name to save TF model.')
args = parser.parse_args()
# load model config
config_path = args.config_path
c = load_config(config_path)
num_speakers = 0
# init torch model
model = setup_generator(c)
checkpoint = torch.load(args.torch_model_path,
map_location=torch.device('cpu'))
state_dict = checkpoint['model']
model.load_state_dict(state_dict)
model.remove_weight_norm()
state_dict = model.state_dict()
# init tf model
model_tf = setup_tf_generator(c)
common_sufix = '/.ATTRIBUTES/VARIABLE_VALUE'
import os
import numpy as np
from torch.utils.data import DataLoader
from TTS.vocoder.datasets.gan_dataset import GANDataset
from TTS.vocoder.datasets.preprocess import load_wav_data
from TTS.utils.audio import AudioProcessor
from TTS.utils.io import load_config
file_path = os.path.dirname(os.path.realpath(__file__))
OUTPATH = os.path.join(file_path, "../../tests/outputs/loader_tests/")
os.makedirs(OUTPATH, exist_ok=True)
C = load_config(os.path.join(file_path, 'test_config.json'))
test_data_path = os.path.join(file_path, "../../tests/data/ljspeech/")
ok_ljspeech = os.path.exists(test_data_path)
def gan_dataset_case(batch_size, seq_len, hop_len, conv_pad, return_segments,
use_noise_augment, use_cache, num_workers):
''' run dataloader with given parameters and check conditions '''
ap = AudioProcessor(**C.audio)
_, train_items = load_wav_data(test_data_path, 10)
dataset = GANDataset(ap,
train_items,
seq_len=seq_len,
hop_len=hop_len,
pad_short=2000,
conv_pad=conv_pad,
return_segments=return_segments,