Merge branch 'wavernn' of https://github.com/SanjaESC/TTS into SanjaESC-wavernn

This commit is contained in:
erogol 2020-10-26 16:50:45 +01:00
commit 741d278cac
17 changed files with 1694 additions and 18 deletions

View File

@ -17,5 +17,6 @@ fi
if [[ "$TEST_SUITE" == "testscripts" ]]; then
# test model training scripts
./tests/test_tts_train.sh
./tests/test_vocoder_train.sh
./tests/test_vocoder_gan_train.sh
./tests/test_vocoder_wavernn_train.sh
fi

View File

@ -11,19 +11,22 @@
<br/>
Mozilla TTS is a deep learning based Text2Speech project, low in cost and high in quality.
This project is a part of [Mozilla Common Voice](https://voice.mozilla.org/en).
Mozilla TTS aims a deep learning based Text2Speech engine, low in cost and high in quality.
English Voice Samples: https://erogol.github.io/ddc-samples/
You can check some of synthesized voice samples from [here](https://erogol.github.io/ddc-samples/).
TTS training recipes: https://github.com/erogol/TTS_recipes
If you are new, you can also find [here](http://www.erogol.com/text-speech-deep-learning-architectures/) a brief post about some of TTS architectures and [here](https://github.com/erogol/TTS-papers) list of up-to-date research papers.
TTS paper collection: https://github.com/erogol/TTS-papers
[![](https://sourcerer.io/fame/erogol/erogol/TTS/images/0)](https://sourcerer.io/fame/erogol/erogol/TTS/links/0)[![](https://sourcerer.io/fame/erogol/erogol/TTS/images/1)](https://sourcerer.io/fame/erogol/erogol/TTS/links/1)[![](https://sourcerer.io/fame/erogol/erogol/TTS/images/2)](https://sourcerer.io/fame/erogol/erogol/TTS/links/2)[![](https://sourcerer.io/fame/erogol/erogol/TTS/images/3)](https://sourcerer.io/fame/erogol/erogol/TTS/links/3)[![](https://sourcerer.io/fame/erogol/erogol/TTS/images/4)](https://sourcerer.io/fame/erogol/erogol/TTS/links/4)[![](https://sourcerer.io/fame/erogol/erogol/TTS/images/5)](https://sourcerer.io/fame/erogol/erogol/TTS/links/5)[![](https://sourcerer.io/fame/erogol/erogol/TTS/images/6)](https://sourcerer.io/fame/erogol/erogol/TTS/links/6)[![](https://sourcerer.io/fame/erogol/erogol/TTS/images/7)](https://sourcerer.io/fame/erogol/erogol/TTS/links/7)
## TTS Performance
<p align="center"><img src="https://camo.githubusercontent.com/9fa79f977015e55eb9ec7aa32045555f60d093d3/68747470733a2f2f646973636f757273652d706161732d70726f64756374696f6e2d636f6e74656e742e73332e6475616c737461636b2e75732d656173742d312e616d617a6f6e6177732e636f6d2f6f7074696d697a65642f33582f362f342f363432386639383065396563373531633234386535393134363038393566373838316165633063365f325f363930783339342e706e67"/></p>
<p align="center"><img src="https://discourse-prod-uploads-81679984178418.s3.dualstack.us-west-2.amazonaws.com/optimized/3X/6/4/6428f980e9ec751c248e591460895f7881aec0c6_2_1035x591.png" width="800" /></p>
"Mozilla*" and "Judy*" are our models.
[Details...](https://github.com/mozilla/TTS/wiki/Mean-Opinion-Score-Results)
## Provided Models and Methods

View File

@ -11,6 +11,7 @@ from TTS.tts.datasets.preprocess import load_meta_data
from TTS.utils.io import load_config
from TTS.utils.audio import AudioProcessor
def main():
"""Run preprocessing process."""
parser = argparse.ArgumentParser(

View File

@ -326,7 +326,6 @@ def evaluate(model_G, criterion_G, model_D, criterion_D, ap, global_step, epoch)
y_hat = model_G.pqmf_synthesis(y_hat)
y_G_sub = model_G.pqmf_analysis(y_G)
scores_fake, feats_fake, feats_real = None, None, None
if global_step > c.steps_to_start_discriminator:
@ -403,7 +402,6 @@ def evaluate(model_G, criterion_G, model_D, criterion_D, ap, global_step, epoch)
else:
loss_dict[key] = value.item()
step_time = time.time() - start_time
epoch_time += step_time
@ -443,7 +441,8 @@ def main(args): # pylint: disable=redefined-outer-name
print(f" > Loading wavs from: {c.data_path}")
if c.feature_path is not None:
print(f" > Loading features from: {c.feature_path}")
eval_data, train_data = load_wav_feat_data(c.data_path, c.feature_path, c.eval_split_size)
eval_data, train_data = load_wav_feat_data(
c.data_path, c.feature_path, c.eval_split_size)
else:
eval_data, train_data = load_wav_data(c.data_path, c.eval_split_size)
@ -470,10 +469,12 @@ def main(args): # pylint: disable=redefined-outer-name
scheduler_disc = None
if 'lr_scheduler_gen' in c:
scheduler_gen = getattr(torch.optim.lr_scheduler, c.lr_scheduler_gen)
scheduler_gen = scheduler_gen(optimizer_gen, **c.lr_scheduler_gen_params)
scheduler_gen = scheduler_gen(
optimizer_gen, **c.lr_scheduler_gen_params)
if 'lr_scheduler_disc' in c:
scheduler_disc = getattr(torch.optim.lr_scheduler, c.lr_scheduler_disc)
scheduler_disc = scheduler_disc(optimizer_disc, **c.lr_scheduler_disc_params)
scheduler_disc = scheduler_disc(
optimizer_disc, **c.lr_scheduler_disc_params)
# setup criterion
criterion_gen = GeneratorLoss(c)
@ -572,8 +573,7 @@ if __name__ == '__main__':
parser.add_argument(
'--continue_path',
type=str,
help=
'Training output folder to continue training. Use to continue a training. If it is used, "config_path" is ignored.',
help='Training output folder to continue training. Use to continue a training. If it is used, "config_path" is ignored.',
default='',
required='--config_path' not in sys.argv)
parser.add_argument(

View File

@ -0,0 +1,515 @@
import argparse
import os
import sys
import traceback
import time
import glob
import random
import torch
from torch.utils.data import DataLoader
# from torch.utils.data.distributed import DistributedSampler
from TTS.tts.utils.visual import plot_spectrogram
from TTS.utils.audio import AudioProcessor
from TTS.utils.radam import RAdam
from TTS.utils.io import copy_config_file, load_config
from TTS.utils.training import setup_torch_training_env
from TTS.utils.console_logger import ConsoleLogger
from TTS.utils.tensorboard_logger import TensorboardLogger
from TTS.utils.generic_utils import (
KeepAverage,
count_parameters,
create_experiment_folder,
get_git_branch,
remove_experiment_folder,
set_init_dict,
)
from TTS.vocoder.datasets.wavernn_dataset import WaveRNNDataset
from TTS.vocoder.datasets.preprocess import (
load_wav_data,
load_wav_feat_data
)
from TTS.vocoder.utils.distribution import discretized_mix_logistic_loss, gaussian_loss
from TTS.vocoder.utils.generic_utils import setup_wavernn
from TTS.vocoder.utils.io import save_best_model, save_checkpoint
use_cuda, num_gpus = setup_torch_training_env(True, True)
def setup_loader(ap, is_val=False, verbose=False):
if is_val and not c.run_eval:
loader = None
else:
dataset = WaveRNNDataset(ap=ap,
items=eval_data if is_val else train_data,
seq_len=c.seq_len,
hop_len=ap.hop_length,
pad=c.padding,
mode=c.mode,
mulaw=c.mulaw,
is_training=not is_val,
verbose=verbose,
)
# sampler = DistributedSampler(dataset) if num_gpus > 1 else None
loader = DataLoader(dataset,
shuffle=True,
collate_fn=dataset.collate,
batch_size=c.batch_size,
num_workers=c.num_val_loader_workers
if is_val
else c.num_loader_workers,
pin_memory=True,
)
return loader
def format_data(data):
# setup input data
x_input = data[0]
mels = data[1]
y_coarse = data[2]
# dispatch data to GPU
if use_cuda:
x_input = x_input.cuda(non_blocking=True)
mels = mels.cuda(non_blocking=True)
y_coarse = y_coarse.cuda(non_blocking=True)
return x_input, mels, y_coarse
def train(model, optimizer, criterion, scheduler, ap, global_step, epoch):
# create train loader
data_loader = setup_loader(ap, is_val=False, verbose=(epoch == 0))
model.train()
epoch_time = 0
keep_avg = KeepAverage()
if use_cuda:
batch_n_iter = int(len(data_loader.dataset) /
(c.batch_size * num_gpus))
else:
batch_n_iter = int(len(data_loader.dataset) / c.batch_size)
end_time = time.time()
c_logger.print_train_start()
# train loop
print(" > Training", flush=True)
for num_iter, data in enumerate(data_loader):
start_time = time.time()
x_input, mels, y_coarse = format_data(data)
loader_time = time.time() - end_time
global_step += 1
y_hat = model(x_input, mels)
if isinstance(model.mode, int):
y_hat = y_hat.transpose(1, 2).unsqueeze(-1)
else:
y_coarse = y_coarse.float()
y_coarse = y_coarse.unsqueeze(-1)
# compute losses
loss = criterion(y_hat, y_coarse)
if loss.item() is None:
raise RuntimeError(" [!] None loss. Exiting ...")
optimizer.zero_grad()
loss.backward()
if c.grad_clip > 0:
torch.nn.utils.clip_grad_norm_(
model.parameters(), c.grad_clip)
optimizer.step()
if scheduler is not None:
scheduler.step()
# get the current learning rate
cur_lr = list(optimizer.param_groups)[0]["lr"]
step_time = time.time() - start_time
epoch_time += step_time
update_train_values = dict()
loss_dict = dict()
loss_dict["model_loss"] = loss.item()
for key, value in loss_dict.items():
update_train_values["avg_" + key] = value
update_train_values["avg_loader_time"] = loader_time
update_train_values["avg_step_time"] = step_time
keep_avg.update_values(update_train_values)
# print training stats
if global_step % c.print_step == 0:
log_dict = {"step_time": [step_time, 2],
"loader_time": [loader_time, 4],
"current_lr": cur_lr,
}
c_logger.print_train_step(batch_n_iter,
num_iter,
global_step,
log_dict,
loss_dict,
keep_avg.avg_values,
)
# plot step stats
if global_step % 10 == 0:
iter_stats = {"lr": cur_lr, "step_time": step_time}
iter_stats.update(loss_dict)
tb_logger.tb_train_iter_stats(global_step, iter_stats)
# save checkpoint
if global_step % c.save_step == 0:
if c.checkpoint:
# save model
save_checkpoint(model,
optimizer,
scheduler,
None,
None,
None,
global_step,
epoch,
OUT_PATH,
model_losses=loss_dict,
)
# synthesize a full voice
rand_idx = random.randrange(0, len(train_data))
wav_path = train_data[rand_idx] if not isinstance(
train_data[rand_idx], (tuple, list)) else train_data[rand_idx][0]
wav = ap.load_wav(wav_path)
ground_mel = ap.melspectrogram(wav)
sample_wav = model.generate(ground_mel,
c.batched,
c.target_samples,
c.overlap_samples,
use_cuda
)
predict_mel = ap.melspectrogram(sample_wav)
# compute spectrograms
figures = {"train/ground_truth": plot_spectrogram(ground_mel.T),
"train/prediction": plot_spectrogram(predict_mel.T)
}
tb_logger.tb_train_figures(global_step, figures)
# Sample audio
tb_logger.tb_train_audios(
global_step, {
"train/audio": sample_wav}, c.audio["sample_rate"]
)
end_time = time.time()
# print epoch stats
c_logger.print_train_epoch_end(global_step, epoch, epoch_time, keep_avg)
# Plot Training Epoch Stats
epoch_stats = {"epoch_time": epoch_time}
epoch_stats.update(keep_avg.avg_values)
tb_logger.tb_train_epoch_stats(global_step, epoch_stats)
# TODO: plot model stats
# if c.tb_model_param_stats:
# tb_logger.tb_model_weights(model, global_step)
return keep_avg.avg_values, global_step
@torch.no_grad()
def evaluate(model, criterion, ap, global_step, epoch):
# create train loader
data_loader = setup_loader(ap, is_val=True, verbose=(epoch == 0))
model.eval()
epoch_time = 0
keep_avg = KeepAverage()
end_time = time.time()
c_logger.print_eval_start()
with torch.no_grad():
for num_iter, data in enumerate(data_loader):
start_time = time.time()
# format data
x_input, mels, y_coarse = format_data(data)
loader_time = time.time() - end_time
global_step += 1
y_hat = model(x_input, mels)
if isinstance(model.mode, int):
y_hat = y_hat.transpose(1, 2).unsqueeze(-1)
else:
y_coarse = y_coarse.float()
y_coarse = y_coarse.unsqueeze(-1)
loss = criterion(y_hat, y_coarse)
# Compute avg loss
# if num_gpus > 1:
# loss = reduce_tensor(loss.data, num_gpus)
loss_dict = dict()
loss_dict["model_loss"] = loss.item()
step_time = time.time() - start_time
epoch_time += step_time
# update avg stats
update_eval_values = dict()
for key, value in loss_dict.items():
update_eval_values["avg_" + key] = value
update_eval_values["avg_loader_time"] = loader_time
update_eval_values["avg_step_time"] = step_time
keep_avg.update_values(update_eval_values)
# print eval stats
if c.print_eval:
c_logger.print_eval_step(
num_iter, loss_dict, keep_avg.avg_values)
if epoch % c.test_every_epochs == 0 and epoch != 0:
# synthesize a full voice
rand_idx = random.randrange(0, len(eval_data))
wav_path = eval_data[rand_idx] if not isinstance(
eval_data[rand_idx], (tuple, list)) else eval_data[rand_idx][0]
wav = ap.load_wav(wav_path)
ground_mel = ap.melspectrogram(wav)
sample_wav = model.generate(ground_mel,
c.batched,
c.target_samples,
c.overlap_samples,
use_cuda
)
predict_mel = ap.melspectrogram(sample_wav)
# Sample audio
tb_logger.tb_eval_audios(
global_step, {
"eval/audio": sample_wav}, c.audio["sample_rate"]
)
# compute spectrograms
figures = {"eval/ground_truth": plot_spectrogram(ground_mel.T),
"eval/prediction": plot_spectrogram(predict_mel.T)
}
tb_logger.tb_eval_figures(global_step, figures)
tb_logger.tb_eval_stats(global_step, keep_avg.avg_values)
return keep_avg.avg_values
# FIXME: move args definition/parsing inside of main?
def main(args): # pylint: disable=redefined-outer-name
# pylint: disable=global-variable-undefined
global train_data, eval_data
# setup audio processor
ap = AudioProcessor(**c.audio)
# print(f" > Loading wavs from: {c.data_path}")
# if c.feature_path is not None:
# print(f" > Loading features from: {c.feature_path}")
# eval_data, train_data = load_wav_feat_data(
# c.data_path, c.feature_path, c.eval_split_size
# )
# else:
# mel_feat_path = os.path.join(OUT_PATH, "mel")
# feat_data = find_feat_files(mel_feat_path)
# if feat_data:
# print(f" > Loading features from: {mel_feat_path}")
# eval_data, train_data = load_wav_feat_data(
# c.data_path, mel_feat_path, c.eval_split_size
# )
# else:
# print(" > No feature data found. Preprocessing...")
# # preprocessing feature data from given wav files
# preprocess_wav_files(OUT_PATH, CONFIG, ap)
# eval_data, train_data = load_wav_feat_data(
# c.data_path, mel_feat_path, c.eval_split_size
# )
print(f" > Loading wavs from: {c.data_path}")
if c.feature_path is not None:
print(f" > Loading features from: {c.feature_path}")
eval_data, train_data = load_wav_feat_data(
c.data_path, c.feature_path, c.eval_split_size)
else:
eval_data, train_data = load_wav_data(
c.data_path, c.eval_split_size)
# setup model
model_wavernn = setup_wavernn(c)
# define train functions
if c.mode == "mold":
criterion = discretized_mix_logistic_loss
elif c.mode == "gauss":
criterion = gaussian_loss
elif isinstance(c.mode, int):
criterion = torch.nn.CrossEntropyLoss()
if use_cuda:
model_wavernn.cuda()
if isinstance(c.mode, int):
criterion.cuda()
optimizer = RAdam(model_wavernn.parameters(), lr=c.lr, weight_decay=0)
scheduler = None
if "lr_scheduler" in c:
scheduler = getattr(torch.optim.lr_scheduler, c.lr_scheduler)
scheduler = scheduler(optimizer, **c.lr_scheduler_params)
# slow start for the first 5 epochs
# lr_lambda = lambda epoch: min(epoch / c.warmup_steps, 1)
# scheduler = optim.lr_scheduler.LambdaLR(optimizer, lr_lambda)
# restore any checkpoint
if args.restore_path:
checkpoint = torch.load(args.restore_path, map_location="cpu")
try:
print(" > Restoring Model...")
model_wavernn.load_state_dict(checkpoint["model"])
print(" > Restoring Optimizer...")
optimizer.load_state_dict(checkpoint["optimizer"])
if "scheduler" in checkpoint:
print(" > Restoring Generator LR Scheduler...")
scheduler.load_state_dict(checkpoint["scheduler"])
scheduler.optimizer = optimizer
# TODO: fix resetting restored optimizer lr
# optimizer.load_state_dict(checkpoint["optimizer"])
except RuntimeError:
# retore only matching layers.
print(" > Partial model initialization...")
model_dict = model_wavernn.state_dict()
model_dict = set_init_dict(model_dict, checkpoint["model"], c)
model_wavernn.load_state_dict(model_dict)
print(" > Model restored from step %d" %
checkpoint["step"], flush=True)
args.restore_step = checkpoint["step"]
else:
args.restore_step = 0
# DISTRIBUTED
# if num_gpus > 1:
# model = apply_gradient_allreduce(model)
num_parameters = count_parameters(model_wavernn)
print(" > Model has {} parameters".format(num_parameters), flush=True)
if "best_loss" not in locals():
best_loss = float("inf")
global_step = args.restore_step
for epoch in range(0, c.epochs):
c_logger.print_epoch_start(epoch, c.epochs)
_, global_step = train(model_wavernn, optimizer,
criterion, scheduler, ap, global_step, epoch)
eval_avg_loss_dict = evaluate(
model_wavernn, criterion, ap, global_step, epoch)
c_logger.print_epoch_end(epoch, eval_avg_loss_dict)
target_loss = eval_avg_loss_dict["avg_model_loss"]
best_loss = save_best_model(
target_loss,
best_loss,
model_wavernn,
optimizer,
scheduler,
None,
None,
None,
global_step,
epoch,
OUT_PATH,
model_losses=eval_avg_loss_dict,
)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument(
"--continue_path",
type=str,
help='Training output folder to continue training. Use to continue a training. If it is used, "config_path" is ignored.',
default="",
required="--config_path" not in sys.argv,
)
parser.add_argument(
"--restore_path",
type=str,
help="Model file to be restored. Use to finetune a model.",
default="",
)
parser.add_argument(
"--config_path",
type=str,
help="Path to config file for training.",
required="--continue_path" not in sys.argv,
)
parser.add_argument(
"--debug",
type=bool,
default=False,
help="Do not verify commit integrity to run training.",
)
# DISTRUBUTED
parser.add_argument(
"--rank",
type=int,
default=0,
help="DISTRIBUTED: process rank for distributed training.",
)
parser.add_argument(
"--group_id", type=str, 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__))
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 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()
copy_config_file(
args.config_path, os.path.join(OUT_PATH, "c.json"), new_fields
)
os.chmod(AUDIO_PATH, 0o775)
os.chmod(OUT_PATH, 0o775)
LOG_DIR = OUT_PATH
tb_logger = TensorboardLogger(LOG_DIR, model_name="VOCODER")
# write model desc to tensorboard
tb_logger.tb_add_text("model-description", c["run_description"], 0)
try:
main(args)
except KeyboardInterrupt:
remove_experiment_folder(OUT_PATH)
try:
sys.exit(0)
except SystemExit:
os._exit(0) # pylint: disable=protected-access
except Exception: # pylint: disable=broad-except
remove_experiment_folder(OUT_PATH)
traceback.print_exc()
sys.exit(1)

View File

@ -0,0 +1,97 @@
{
"run_name": "wavernn_test",
"run_description": "wavernn_test training",
// AUDIO PARAMETERS
"audio": {
"fft_size": 1024, // number of stft frequency levels. Size of the linear spectogram frame.
"win_length": 1024, // stft window length in ms.
"hop_length": 256, // stft window hop-lengh in ms.
"frame_length_ms": null, // stft window length in ms.If null, 'win_length' is used.
"frame_shift_ms": null, // stft window hop-lengh in ms. If null, 'hop_length' is used.
// Audio processing parameters
"sample_rate": 22050, // DATASET-RELATED: wav sample-rate. If different than the original data, it is resampled.
"preemphasis": 0.98, // pre-emphasis to reduce spec noise and make it more structured. If 0.0, no -pre-emphasis.
"ref_level_db": 20, // reference level db, theoretically 20db is the sound of air.
// Silence trimming
"do_trim_silence": false, // enable trimming of slience of audio as you load it. LJspeech (false), TWEB (false), Nancy (true)
"trim_db": 60, // threshold for timming silence. Set this according to your dataset.
// MelSpectrogram parameters
"num_mels": 80, // size of the mel spec frame.
"mel_fmin": 40.0, // minimum freq level for mel-spec. ~50 for male and ~95 for female voices. Tune for dataset!!
"mel_fmax": 8000.0, // maximum freq level for mel-spec. Tune for dataset!!
"spec_gain": 20.0, // scaler value appplied after log transform of spectrogram.
// Normalization parameters
"signal_norm": true, // normalize spec values. Mean-Var normalization if 'stats_path' is defined otherwise range normalization defined by the other params.
"min_level_db": -100, // lower bound for normalization
"symmetric_norm": true, // move normalization to range [-1, 1]
"max_norm": 4.0, // scale normalization to range [-max_norm, max_norm] or [0, max_norm]
"clip_norm": true, // clip normalized values into the range.
"stats_path": null // DO NOT USE WITH MULTI_SPEAKER MODEL. scaler stats file computed by 'compute_statistics.py'. If it is defined, mean-std based notmalization is used and other normalization params are ignored
},
// Generating / Synthesizing
"batched": true,
"target_samples": 11000, // target number of samples to be generated in each batch entry
"overlap_samples": 550, // number of samples for crossfading between batches
// DISTRIBUTED TRAINING
// "distributed":{
// "backend": "nccl",
// "url": "tcp:\/\/localhost:54321"
// },
// MODEL MODE
"mode": 10, // mold [string], gauss [string], bits [int]
"mulaw": true, // apply mulaw if mode is bits
// MODEL PARAMETERS
"wavernn_model_params": {
"rnn_dims": 512,
"fc_dims": 512,
"compute_dims": 128,
"res_out_dims": 128,
"num_res_blocks": 10,
"use_aux_net": true,
"use_upsample_net": true,
"upsample_factors": [4, 8, 8] // this needs to correctly factorise hop_length
},
// DATASET
//"use_gta": true, // use computed gta features from the tts model
"data_path": "/media/alexander/LinuxFS/SpeechData/GothicSpeech/NPC_Speech", // path containing training wav files
"feature_path": null, // path containing computed features from wav files if null compute them
"seq_len": 1280, // has to be devideable by hop_length
"padding": 2, // pad the input for resnet to see wider input length
// TRAINING
"batch_size": 64, // Batch size for training.
"epochs": 10000, // total number of epochs to train.
// VALIDATION
"run_eval": true,
"test_every_epochs": 10, // Test after set number of epochs (Test every 10 epochs for example)
// OPTIMIZER
"grad_clip": 4, // apply gradient clipping if > 0
"lr_scheduler": "MultiStepLR", // one of the schedulers from https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate
"lr_scheduler_params": {
"gamma": 0.5,
"milestones": [200000, 400000, 600000]
},
"lr": 1e-4, // initial learning rate
// TENSORBOARD and LOGGING
"print_step": 25, // Number of steps to log traning on console.
"print_eval": false, // If True, it prints loss values for each step in eval run.
"save_step": 25000, // Number of training steps expected to plot training stats on TB and save model checkpoints.
"checkpoint": true, // If true, it saves checkpoints per "save_step"
"tb_model_param_stats": false, // true, plots param stats per layer on tensorboard. Might be memory consuming, but good for debugging.
// DATA LOADING
"num_loader_workers": 4, // number of training data loader processes. Don't set it too big. 4-8 are good values.
"num_val_loader_workers": 4, // number of evaluation data loader processes.
"eval_split_size": 50, // number of samples for testing
// PATHS
"output_path": "output/training/path"
}

View File

@ -1,17 +1,38 @@
import glob
import os
from pathlib import Path
from tqdm import tqdm
import numpy as np
def preprocess_wav_files(out_path, config, ap):
os.makedirs(os.path.join(out_path, "quant"), exist_ok=True)
os.makedirs(os.path.join(out_path, "mel"), exist_ok=True)
wav_files = find_wav_files(config.data_path)
for path in tqdm(wav_files):
wav_name = Path(path).stem
quant_path = os.path.join(out_path, "quant", wav_name + ".npy")
mel_path = os.path.join(out_path, "mel", wav_name + ".npy")
y = ap.load_wav(path)
mel = ap.melspectrogram(y)
np.save(mel_path, mel)
if isinstance(config.mode, int):
quant = (
ap.mulaw_encode(y, qc=config.mode)
if config.mulaw
else ap.quantize(y, bits=config.mode)
)
np.save(quant_path, quant)
def find_wav_files(data_path):
wav_paths = glob.glob(os.path.join(data_path, '**', '*.wav'), recursive=True)
wav_paths = glob.glob(os.path.join(data_path, "**", "*.wav"), recursive=True)
return wav_paths
def find_feat_files(data_path):
feat_paths = glob.glob(os.path.join(data_path, '**', '*.npy'), recursive=True)
feat_paths = glob.glob(os.path.join(data_path, "**", "*.npy"), recursive=True)
return feat_paths
@ -23,8 +44,12 @@ def load_wav_data(data_path, eval_split_size):
def load_wav_feat_data(data_path, feat_path, eval_split_size):
wav_paths = sorted(find_wav_files(data_path))
feat_paths = sorted(find_feat_files(feat_path))
wav_paths = find_wav_files(data_path)
feat_paths = find_feat_files(feat_path)
wav_paths.sort(key=lambda x: Path(x).stem)
feat_paths.sort(key=lambda x: Path(x).stem)
assert len(wav_paths) == len(feat_paths)
for wav, feat in zip(wav_paths, feat_paths):
wav_name = Path(wav).stem

View File

@ -0,0 +1,115 @@
import torch
import numpy as np
from torch.utils.data import Dataset
class WaveRNNDataset(Dataset):
"""
WaveRNN Dataset searchs for all the wav files under root path
and converts them to acoustic features on the fly.
"""
def __init__(self,
ap,
items,
seq_len,
hop_len,
pad,
mode,
mulaw,
is_training=True,
verbose=False,
):
self.ap = ap
self.compute_feat = not isinstance(items[0], (tuple, list))
self.item_list = items
self.seq_len = seq_len
self.hop_len = hop_len
self.pad = pad
self.mode = mode
self.mulaw = mulaw
self.is_training = is_training
self.verbose = verbose
def __len__(self):
return len(self.item_list)
def __getitem__(self, index):
item = self.load_item(index)
return item
def load_item(self, index):
"""
load (audio, feat) couple if feature_path is set
else compute it on the fly
"""
if self.compute_feat:
wavpath = self.item_list[index]
audio = self.ap.load_wav(wavpath)
mel = self.ap.melspectrogram(audio)
if mel.shape[-1] < 5:
print(" [!] Instance is too short! : {}".format(wavpath))
self.item_list[index] = self.item_list[index + 1]
audio = self.ap.load_wav(wavpath)
mel = self.ap.melspectrogram(audio)
if self.mode in ["gauss", "mold"]:
x_input = audio
elif isinstance(self.mode, int):
x_input = (self.ap.mulaw_encode(audio, qc=self.mode)
if self.mulaw else self.ap.quantize(audio, bits=self.mode))
else:
raise RuntimeError("Unknown dataset mode - ", self.mode)
else:
wavpath, feat_path = self.item_list[index]
mel = np.load(feat_path.replace("/quant/", "/mel/"))
if mel.shape[-1] < 5:
print(" [!] Instance is too short! : {}".format(wavpath))
self.item_list[index] = self.item_list[index + 1]
feat_path = self.item_list[index]
mel = np.load(feat_path.replace("/quant/", "/mel/"))
if self.mode in ["gauss", "mold"]:
x_input = self.ap.load_wav(wavpath)
elif isinstance(self.mode, int):
x_input = np.load(feat_path.replace("/mel/", "/quant/"))
else:
raise RuntimeError("Unknown dataset mode - ", self.mode)
return mel, x_input
def collate(self, batch):
mel_win = self.seq_len // self.hop_len + 2 * self.pad
max_offsets = [x[0].shape[-1] -
(mel_win + 2 * self.pad) for x in batch]
mel_offsets = [np.random.randint(0, offset) for offset in max_offsets]
sig_offsets = [(offset + self.pad) *
self.hop_len for offset in mel_offsets]
mels = [
x[0][:, mel_offsets[i]: mel_offsets[i] + mel_win]
for i, x in enumerate(batch)
]
coarse = [
x[1][sig_offsets[i]: sig_offsets[i] + self.seq_len + 1]
for i, x in enumerate(batch)
]
mels = np.stack(mels).astype(np.float32)
if self.mode in ["gauss", "mold"]:
coarse = np.stack(coarse).astype(np.float32)
coarse = torch.FloatTensor(coarse)
x_input = coarse[:, : self.seq_len]
elif isinstance(self.mode, int):
coarse = np.stack(coarse).astype(np.int64)
coarse = torch.LongTensor(coarse)
x_input = (2 * coarse[:, : self.seq_len].float() /
(2 ** self.mode - 1.0) - 1.0)
y_coarse = coarse[:, 1:]
mels = torch.FloatTensor(mels)
return x_input, mels, y_coarse

View File

@ -0,0 +1,496 @@
import sys
import torch
import torch.nn as nn
import numpy as np
import torch.nn.functional as F
import time
# fix this
from TTS.utils.audio import AudioProcessor as ap
from TTS.vocoder.utils.distribution import (
sample_from_gaussian,
sample_from_discretized_mix_logistic,
)
def stream(string, variables):
sys.stdout.write(f"\r{string}" % variables)
class ResBlock(nn.Module):
def __init__(self, dims):
super().__init__()
self.conv1 = nn.Conv1d(dims, dims, kernel_size=1, bias=False)
self.conv2 = nn.Conv1d(dims, dims, kernel_size=1, bias=False)
self.batch_norm1 = nn.BatchNorm1d(dims)
self.batch_norm2 = nn.BatchNorm1d(dims)
def forward(self, x):
residual = x
x = self.conv1(x)
x = self.batch_norm1(x)
x = F.relu(x)
x = self.conv2(x)
x = self.batch_norm2(x)
return x + residual
class MelResNet(nn.Module):
def __init__(self, num_res_blocks, in_dims, compute_dims, res_out_dims, pad):
super().__init__()
k_size = pad * 2 + 1
self.conv_in = nn.Conv1d(
in_dims, compute_dims, kernel_size=k_size, bias=False)
self.batch_norm = nn.BatchNorm1d(compute_dims)
self.layers = nn.ModuleList()
for _ in range(num_res_blocks):
self.layers.append(ResBlock(compute_dims))
self.conv_out = nn.Conv1d(compute_dims, res_out_dims, kernel_size=1)
def forward(self, x):
x = self.conv_in(x)
x = self.batch_norm(x)
x = F.relu(x)
for f in self.layers:
x = f(x)
x = self.conv_out(x)
return x
class Stretch2d(nn.Module):
def __init__(self, x_scale, y_scale):
super().__init__()
self.x_scale = x_scale
self.y_scale = y_scale
def forward(self, x):
b, c, h, w = x.size()
x = x.unsqueeze(-1).unsqueeze(3)
x = x.repeat(1, 1, 1, self.y_scale, 1, self.x_scale)
return x.view(b, c, h * self.y_scale, w * self.x_scale)
class UpsampleNetwork(nn.Module):
def __init__(
self,
feat_dims,
upsample_scales,
compute_dims,
num_res_blocks,
res_out_dims,
pad,
use_aux_net,
):
super().__init__()
self.total_scale = np.cumproduct(upsample_scales)[-1]
self.indent = pad * self.total_scale
self.use_aux_net = use_aux_net
if use_aux_net:
self.resnet = MelResNet(
num_res_blocks, feat_dims, compute_dims, res_out_dims, pad
)
self.resnet_stretch = Stretch2d(self.total_scale, 1)
self.up_layers = nn.ModuleList()
for scale in upsample_scales:
k_size = (1, scale * 2 + 1)
padding = (0, scale)
stretch = Stretch2d(scale, 1)
conv = nn.Conv2d(1, 1, kernel_size=k_size,
padding=padding, bias=False)
conv.weight.data.fill_(1.0 / k_size[1])
self.up_layers.append(stretch)
self.up_layers.append(conv)
def forward(self, m):
if self.use_aux_net:
aux = self.resnet(m).unsqueeze(1)
aux = self.resnet_stretch(aux)
aux = aux.squeeze(1)
aux = aux.transpose(1, 2)
else:
aux = None
m = m.unsqueeze(1)
for f in self.up_layers:
m = f(m)
m = m.squeeze(1)[:, :, self.indent: -self.indent]
return m.transpose(1, 2), aux
class Upsample(nn.Module):
def __init__(
self, scale, pad, num_res_blocks, feat_dims, compute_dims, res_out_dims, use_aux_net
):
super().__init__()
self.scale = scale
self.pad = pad
self.indent = pad * scale
self.use_aux_net = use_aux_net
self.resnet = MelResNet(num_res_blocks, feat_dims,
compute_dims, res_out_dims, pad)
def forward(self, m):
if self.use_aux_net:
aux = self.resnet(m)
aux = torch.nn.functional.interpolate(
aux, scale_factor=self.scale, mode="linear", align_corners=True
)
aux = aux.transpose(1, 2)
else:
aux = None
m = torch.nn.functional.interpolate(
m, scale_factor=self.scale, mode="linear", align_corners=True
)
m = m[:, :, self.indent: -self.indent]
m = m * 0.045 # empirically found
return m.transpose(1, 2), aux
class WaveRNN(nn.Module):
def __init__(self,
rnn_dims,
fc_dims,
mode,
mulaw,
pad,
use_aux_net,
use_upsample_net,
upsample_factors,
feat_dims,
compute_dims,
res_out_dims,
num_res_blocks,
hop_length,
sample_rate,
):
super().__init__()
self.mode = mode
self.mulaw = mulaw
self.pad = pad
self.use_upsample_net = use_upsample_net
self.use_aux_net = use_aux_net
if isinstance(self.mode, int):
self.n_classes = 2 ** self.mode
elif self.mode == "mold":
self.n_classes = 3 * 10
elif self.mode == "gauss":
self.n_classes = 2
else:
raise RuntimeError("Unknown model mode value - ", self.mode)
self.rnn_dims = rnn_dims
self.aux_dims = res_out_dims // 4
self.hop_length = hop_length
self.sample_rate = sample_rate
if self.use_upsample_net:
assert (
np.cumproduct(upsample_factors)[-1] == self.hop_length
), " [!] upsample scales needs to be equal to hop_length"
self.upsample = UpsampleNetwork(
feat_dims,
upsample_factors,
compute_dims,
num_res_blocks,
res_out_dims,
pad,
use_aux_net,
)
else:
self.upsample = Upsample(
hop_length,
pad,
num_res_blocks,
feat_dims,
compute_dims,
res_out_dims,
use_aux_net,
)
if self.use_aux_net:
self.I = nn.Linear(feat_dims + self.aux_dims + 1, rnn_dims)
self.rnn1 = nn.GRU(rnn_dims, rnn_dims, batch_first=True)
self.rnn2 = nn.GRU(rnn_dims + self.aux_dims,
rnn_dims, batch_first=True)
self.fc1 = nn.Linear(rnn_dims + self.aux_dims, fc_dims)
self.fc2 = nn.Linear(fc_dims + self.aux_dims, fc_dims)
self.fc3 = nn.Linear(fc_dims, self.n_classes)
else:
self.I = nn.Linear(feat_dims + 1, rnn_dims)
self.rnn1 = nn.GRU(rnn_dims, rnn_dims, batch_first=True)
self.rnn2 = nn.GRU(rnn_dims, rnn_dims, batch_first=True)
self.fc1 = nn.Linear(rnn_dims, fc_dims)
self.fc2 = nn.Linear(fc_dims, fc_dims)
self.fc3 = nn.Linear(fc_dims, self.n_classes)
def forward(self, x, mels):
bsize = x.size(0)
h1 = torch.zeros(1, bsize, self.rnn_dims).to(x.device)
h2 = torch.zeros(1, bsize, self.rnn_dims).to(x.device)
mels, aux = self.upsample(mels)
if self.use_aux_net:
aux_idx = [self.aux_dims * i for i in range(5)]
a1 = aux[:, :, aux_idx[0]: aux_idx[1]]
a2 = aux[:, :, aux_idx[1]: aux_idx[2]]
a3 = aux[:, :, aux_idx[2]: aux_idx[3]]
a4 = aux[:, :, aux_idx[3]: aux_idx[4]]
x = (
torch.cat([x.unsqueeze(-1), mels, a1], dim=2)
if self.use_aux_net
else torch.cat([x.unsqueeze(-1), mels], dim=2)
)
x = self.I(x)
res = x
self.rnn1.flatten_parameters()
x, _ = self.rnn1(x, h1)
x = x + res
res = x
x = torch.cat([x, a2], dim=2) if self.use_aux_net else x
self.rnn2.flatten_parameters()
x, _ = self.rnn2(x, h2)
x = x + res
x = torch.cat([x, a3], dim=2) if self.use_aux_net else x
x = F.relu(self.fc1(x))
x = torch.cat([x, a4], dim=2) if self.use_aux_net else x
x = F.relu(self.fc2(x))
return self.fc3(x)
def generate(self, mels, batched, target, overlap, use_cuda=False):
self.eval()
device = 'cuda' if use_cuda else 'cpu'
output = []
start = time.time()
rnn1 = self.get_gru_cell(self.rnn1)
rnn2 = self.get_gru_cell(self.rnn2)
with torch.no_grad():
mels = torch.FloatTensor(mels).unsqueeze(0).to(device)
#mels = torch.FloatTensor(mels).cuda().unsqueeze(0)
wave_len = (mels.size(-1) - 1) * self.hop_length
mels = self.pad_tensor(mels.transpose(
1, 2), pad=self.pad, side="both")
mels, aux = self.upsample(mels.transpose(1, 2))
if batched:
mels = self.fold_with_overlap(mels, target, overlap)
if aux is not None:
aux = self.fold_with_overlap(aux, target, overlap)
b_size, seq_len, _ = mels.size()
h1 = torch.zeros(b_size, self.rnn_dims).to(device)
h2 = torch.zeros(b_size, self.rnn_dims).to(device)
x = torch.zeros(b_size, 1).to(device)
if self.use_aux_net:
d = self.aux_dims
aux_split = [aux[:, :, d * i: d * (i + 1)] for i in range(4)]
for i in range(seq_len):
m_t = mels[:, i, :]
if self.use_aux_net:
a1_t, a2_t, a3_t, a4_t = (a[:, i, :] for a in aux_split)
x = (
torch.cat([x, m_t, a1_t], dim=1)
if self.use_aux_net
else torch.cat([x, m_t], dim=1)
)
x = self.I(x)
h1 = rnn1(x, h1)
x = x + h1
inp = torch.cat([x, a2_t], dim=1) if self.use_aux_net else x
h2 = rnn2(inp, h2)
x = x + h2
x = torch.cat([x, a3_t], dim=1) if self.use_aux_net else x
x = F.relu(self.fc1(x))
x = torch.cat([x, a4_t], dim=1) if self.use_aux_net else x
x = F.relu(self.fc2(x))
logits = self.fc3(x)
if self.mode == "mold":
sample = sample_from_discretized_mix_logistic(
logits.unsqueeze(0).transpose(1, 2)
)
output.append(sample.view(-1))
x = sample.transpose(0, 1).to(device)
elif self.mode == "gauss":
sample = sample_from_gaussian(
logits.unsqueeze(0).transpose(1, 2))
output.append(sample.view(-1))
x = sample.transpose(0, 1).to(device)
elif isinstance(self.mode, int):
posterior = F.softmax(logits, dim=1)
distrib = torch.distributions.Categorical(posterior)
sample = 2 * distrib.sample().float() / (self.n_classes - 1.0) - 1.0
output.append(sample)
x = sample.unsqueeze(-1)
else:
raise RuntimeError(
"Unknown model mode value - ", self.mode)
if i % 100 == 0:
self.gen_display(i, seq_len, b_size, start)
output = torch.stack(output).transpose(0, 1)
output = output.cpu().numpy()
output = output.astype(np.float64)
if batched:
output = self.xfade_and_unfold(output, target, overlap)
else:
output = output[0]
if self.mulaw and isinstance(self.mode, int):
output = ap.mulaw_decode(output, self.mode)
# Fade-out at the end to avoid signal cutting out suddenly
fade_out = np.linspace(1, 0, 20 * self.hop_length)
output = output[:wave_len]
if wave_len > len(fade_out):
output[-20 * self.hop_length:] *= fade_out
self.train()
return output
def gen_display(self, i, seq_len, b_size, start):
gen_rate = (i + 1) / (time.time() - start) * b_size / 1000
realtime_ratio = gen_rate * 1000 / self.sample_rate
stream(
"%i/%i -- batch_size: %i -- gen_rate: %.1f kHz -- x_realtime: %.1f ",
(i * b_size, seq_len * b_size, b_size, gen_rate, realtime_ratio),
)
def fold_with_overlap(self, x, target, overlap):
"""Fold the tensor with overlap for quick batched inference.
Overlap will be used for crossfading in xfade_and_unfold()
Args:
x (tensor) : Upsampled conditioning features.
shape=(1, timesteps, features)
target (int) : Target timesteps for each index of batch
overlap (int) : Timesteps for both xfade and rnn warmup
Return:
(tensor) : shape=(num_folds, target + 2 * overlap, features)
Details:
x = [[h1, h2, ... hn]]
Where each h is a vector of conditioning features
Eg: target=2, overlap=1 with x.size(1)=10
folded = [[h1, h2, h3, h4],
[h4, h5, h6, h7],
[h7, h8, h9, h10]]
"""
_, total_len, features = x.size()
# Calculate variables needed
num_folds = (total_len - overlap) // (target + overlap)
extended_len = num_folds * (overlap + target) + overlap
remaining = total_len - extended_len
# Pad if some time steps poking out
if remaining != 0:
num_folds += 1
padding = target + 2 * overlap - remaining
x = self.pad_tensor(x, padding, side="after")
folded = torch.zeros(num_folds, target + 2 *
overlap, features).to(x.device)
# Get the values for the folded tensor
for i in range(num_folds):
start = i * (target + overlap)
end = start + target + 2 * overlap
folded[i] = x[:, start:end, :]
return folded
@staticmethod
def get_gru_cell(gru):
gru_cell = nn.GRUCell(gru.input_size, gru.hidden_size)
gru_cell.weight_hh.data = gru.weight_hh_l0.data
gru_cell.weight_ih.data = gru.weight_ih_l0.data
gru_cell.bias_hh.data = gru.bias_hh_l0.data
gru_cell.bias_ih.data = gru.bias_ih_l0.data
return gru_cell
@staticmethod
def pad_tensor(x, pad, side="both"):
# NB - this is just a quick method i need right now
# i.e., it won't generalise to other shapes/dims
b, t, c = x.size()
total = t + 2 * pad if side == "both" else t + pad
padded = torch.zeros(b, total, c).to(x.device)
if side in ("before", "both"):
padded[:, pad: pad + t, :] = x
elif side == "after":
padded[:, :t, :] = x
return padded
@staticmethod
def xfade_and_unfold(y, target, overlap):
"""Applies a crossfade and unfolds into a 1d array.
Args:
y (ndarry) : Batched sequences of audio samples
shape=(num_folds, target + 2 * overlap)
dtype=np.float64
overlap (int) : Timesteps for both xfade and rnn warmup
Return:
(ndarry) : audio samples in a 1d array
shape=(total_len)
dtype=np.float64
Details:
y = [[seq1],
[seq2],
[seq3]]
Apply a gain envelope at both ends of the sequences
y = [[seq1_in, seq1_target, seq1_out],
[seq2_in, seq2_target, seq2_out],
[seq3_in, seq3_target, seq3_out]]
Stagger and add up the groups of samples:
[seq1_in, seq1_target, (seq1_out + seq2_in), seq2_target, ...]
"""
num_folds, length = y.shape
target = length - 2 * overlap
total_len = num_folds * (target + overlap) + overlap
# Need some silence for the rnn warmup
silence_len = overlap // 2
fade_len = overlap - silence_len
silence = np.zeros((silence_len), dtype=np.float64)
# Equal power crossfade
t = np.linspace(-1, 1, fade_len, dtype=np.float64)
fade_in = np.sqrt(0.5 * (1 + t))
fade_out = np.sqrt(0.5 * (1 - t))
# Concat the silence to the fades
fade_in = np.concatenate([silence, fade_in])
fade_out = np.concatenate([fade_out, silence])
# Apply the gain to the overlap samples
y[:, :overlap] *= fade_in
y[:, -overlap:] *= fade_out
unfolded = np.zeros((total_len), dtype=np.float64)
# Loop to add up all the samples
for i in range(num_folds):
start = i * (target + overlap)
end = start + target + 2 * overlap
unfolded[start:end] += y[i]
return unfolded

View File

@ -0,0 +1,168 @@
import numpy as np
import math
import torch
from torch.distributions.normal import Normal
import torch.nn.functional as F
def gaussian_loss(y_hat, y, log_std_min=-7.0):
assert y_hat.dim() == 3
assert y_hat.size(2) == 2
mean = y_hat[:, :, :1]
log_std = torch.clamp(y_hat[:, :, 1:], min=log_std_min)
# TODO: replace with pytorch dist
log_probs = -0.5 * (
-math.log(2.0 * math.pi)
- 2.0 * log_std
- torch.pow(y - mean, 2) * torch.exp((-2.0 * log_std))
)
return log_probs.squeeze().mean()
def sample_from_gaussian(y_hat, log_std_min=-7.0, scale_factor=1.0):
assert y_hat.size(2) == 2
mean = y_hat[:, :, :1]
log_std = torch.clamp(y_hat[:, :, 1:], min=log_std_min)
dist = Normal(
mean,
torch.exp(log_std),
)
sample = dist.sample()
sample = torch.clamp(torch.clamp(
sample, min=-scale_factor), max=scale_factor)
del dist
return sample
def log_sum_exp(x):
""" numerically stable log_sum_exp implementation that prevents overflow """
# TF ordering
axis = len(x.size()) - 1
m, _ = torch.max(x, dim=axis)
m2, _ = torch.max(x, dim=axis, keepdim=True)
return m + torch.log(torch.sum(torch.exp(x - m2), dim=axis))
# It is adapted from https://github.com/r9y9/wavenet_vocoder/blob/master/wavenet_vocoder/mixture.py
def discretized_mix_logistic_loss(
y_hat, y, num_classes=65536, log_scale_min=None, reduce=True
):
if log_scale_min is None:
log_scale_min = float(np.log(1e-14))
y_hat = y_hat.permute(0, 2, 1)
assert y_hat.dim() == 3
assert y_hat.size(1) % 3 == 0
nr_mix = y_hat.size(1) // 3
# (B x T x C)
y_hat = y_hat.transpose(1, 2)
# unpack parameters. (B, T, num_mixtures) x 3
logit_probs = y_hat[:, :, :nr_mix]
means = y_hat[:, :, nr_mix: 2 * nr_mix]
log_scales = torch.clamp(
y_hat[:, :, 2 * nr_mix: 3 * nr_mix], min=log_scale_min)
# B x T x 1 -> B x T x num_mixtures
y = y.expand_as(means)
centered_y = y - means
inv_stdv = torch.exp(-log_scales)
plus_in = inv_stdv * (centered_y + 1.0 / (num_classes - 1))
cdf_plus = torch.sigmoid(plus_in)
min_in = inv_stdv * (centered_y - 1.0 / (num_classes - 1))
cdf_min = torch.sigmoid(min_in)
# log probability for edge case of 0 (before scaling)
# equivalent: torch.log(F.sigmoid(plus_in))
log_cdf_plus = plus_in - F.softplus(plus_in)
# log probability for edge case of 255 (before scaling)
# equivalent: (1 - F.sigmoid(min_in)).log()
log_one_minus_cdf_min = -F.softplus(min_in)
# probability for all other cases
cdf_delta = cdf_plus - cdf_min
mid_in = inv_stdv * centered_y
# log probability in the center of the bin, to be used in extreme cases
# (not actually used in our code)
log_pdf_mid = mid_in - log_scales - 2.0 * F.softplus(mid_in)
# tf equivalent
# log_probs = tf.where(x < -0.999, log_cdf_plus,
# tf.where(x > 0.999, log_one_minus_cdf_min,
# tf.where(cdf_delta > 1e-5,
# tf.log(tf.maximum(cdf_delta, 1e-12)),
# log_pdf_mid - np.log(127.5))))
# TODO: cdf_delta <= 1e-5 actually can happen. How can we choose the value
# for num_classes=65536 case? 1e-7? not sure..
inner_inner_cond = (cdf_delta > 1e-5).float()
inner_inner_out = inner_inner_cond * torch.log(
torch.clamp(cdf_delta, min=1e-12)
) + (1.0 - inner_inner_cond) * (log_pdf_mid - np.log((num_classes - 1) / 2))
inner_cond = (y > 0.999).float()
inner_out = (
inner_cond * log_one_minus_cdf_min +
(1.0 - inner_cond) * inner_inner_out
)
cond = (y < -0.999).float()
log_probs = cond * log_cdf_plus + (1.0 - cond) * inner_out
log_probs = log_probs + F.log_softmax(logit_probs, -1)
if reduce:
return -torch.mean(log_sum_exp(log_probs))
return -log_sum_exp(log_probs).unsqueeze(-1)
def sample_from_discretized_mix_logistic(y, log_scale_min=None):
"""
Sample from discretized mixture of logistic distributions
Args:
y (Tensor): B x C x T
log_scale_min (float): Log scale minimum value
Returns:
Tensor: sample in range of [-1, 1].
"""
if log_scale_min is None:
log_scale_min = float(np.log(1e-14))
assert y.size(1) % 3 == 0
nr_mix = y.size(1) // 3
# B x T x C
y = y.transpose(1, 2)
logit_probs = y[:, :, :nr_mix]
# sample mixture indicator from softmax
temp = logit_probs.data.new(logit_probs.size()).uniform_(1e-5, 1.0 - 1e-5)
temp = logit_probs.data - torch.log(-torch.log(temp))
_, argmax = temp.max(dim=-1)
# (B, T) -> (B, T, nr_mix)
one_hot = to_one_hot(argmax, nr_mix)
# select logistic parameters
means = torch.sum(y[:, :, nr_mix: 2 * nr_mix] * one_hot, dim=-1)
log_scales = torch.clamp(
torch.sum(y[:, :, 2 * nr_mix: 3 * nr_mix] * one_hot, dim=-1), min=log_scale_min
)
# sample from logistic & clip to interval
# we don't actually round to the nearest 8bit value when sampling
u = means.data.new(means.size()).uniform_(1e-5, 1.0 - 1e-5)
x = means + torch.exp(log_scales) * (torch.log(u) - torch.log(1.0 - u))
x = torch.clamp(torch.clamp(x, min=-1.0), max=1.0)
return x
def to_one_hot(tensor, n, fill_with=1.0):
# we perform one hot encore with respect to the last axis
one_hot = torch.FloatTensor(tensor.size() + (n,)).zero_()
if tensor.is_cuda:
one_hot = one_hot.cuda()
one_hot.scatter_(len(tensor.size()), tensor.unsqueeze(-1), fill_with)
return one_hot

View File

@ -42,6 +42,29 @@ def to_camel(text):
return re.sub(r'(?!^)_([a-zA-Z])', lambda m: m.group(1).upper(), text)
def setup_wavernn(c):
print(" > Model: WaveRNN")
MyModel = importlib.import_module("TTS.vocoder.models.wavernn")
MyModel = getattr(MyModel, "WaveRNN")
model = MyModel(
rnn_dims=c.wavernn_model_params['rnn_dims'],
fc_dims=c.wavernn_model_params['fc_dims'],
mode=c.mode,
mulaw=c.mulaw,
pad=c.padding,
use_aux_net=c.wavernn_model_params['use_aux_net'],
use_upsample_net=c.wavernn_model_params['use_upsample_net'],
upsample_factors=c.wavernn_model_params['upsample_factors'],
feat_dims=c.audio['num_mels'],
compute_dims=c.wavernn_model_params['compute_dims'],
res_out_dims=c.wavernn_model_params['res_out_dims'],
num_res_blocks=c.wavernn_model_params['num_res_blocks'],
hop_length=c.audio["hop_length"],
sample_rate=c.audio["sample_rate"],
)
return model
def setup_generator(c):
print(" > Generator Model: {}".format(c.generator_model))
MyModel = importlib.import_module('TTS.vocoder.models.' +

View File

@ -0,0 +1,94 @@
{
"run_name": "wavernn_test",
"run_description": "wavernn_test training",
// AUDIO PARAMETERS
"audio":{
"fft_size": 1024, // number of stft frequency levels. Size of the linear spectogram frame.
"win_length": 1024, // stft window length in ms.
"hop_length": 256, // stft window hop-lengh in ms.
"frame_length_ms": null, // stft window length in ms.If null, 'win_length' is used.
"frame_shift_ms": null, // stft window hop-lengh in ms. If null, 'hop_length' is used.
// Audio processing parameters
"sample_rate": 22050, // DATASET-RELATED: wav sample-rate. If different than the original data, it is resampled.
"preemphasis": 0.0, // pre-emphasis to reduce spec noise and make it more structured. If 0.0, no -pre-emphasis.
"ref_level_db": 0, // reference level db, theoretically 20db is the sound of air.
// Silence trimming
"do_trim_silence": true,// enable trimming of slience of audio as you load it. LJspeech (false), TWEB (false), Nancy (true)
"trim_db": 60, // threshold for timming silence. Set this according to your dataset.
// MelSpectrogram parameters
"num_mels": 80, // size of the mel spec frame.
"mel_fmin": 0.0, // minimum freq level for mel-spec. ~50 for male and ~95 for female voices. Tune for dataset!!
"mel_fmax": 8000.0, // maximum freq level for mel-spec. Tune for dataset!!
"spec_gain": 20.0, // scaler value appplied after log transform of spectrogram.
// Normalization parameters
"signal_norm": true, // normalize spec values. Mean-Var normalization if 'stats_path' is defined otherwise range normalization defined by the other params.
"min_level_db": -100, // lower bound for normalization
"symmetric_norm": true, // move normalization to range [-1, 1]
"max_norm": 4.0, // scale normalization to range [-max_norm, max_norm] or [0, max_norm]
"clip_norm": true, // clip normalized values into the range.
"stats_path": null // DO NOT USE WITH MULTI_SPEAKER MODEL. scaler stats file computed by 'compute_statistics.py'. If it is defined, mean-std based notmalization is used and other normalization params are ignored
},
// Generating / Synthesizing
"batched": true,
"target_samples": 11000, // target number of samples to be generated in each batch entry
"overlap_samples": 550, // number of samples for crossfading between batches
// DISTRIBUTED TRAINING
// "distributed":{
// "backend": "nccl",
// "url": "tcp:\/\/localhost:54321"
// },
// MODEL PARAMETERS
"use_aux_net": true,
"use_upsample_net": true,
"upsample_factors": [4, 8, 8], // this needs to correctly factorise hop_length
"seq_len": 1280, // has to be devideable by hop_length
"mode": "mold", // mold [string], gauss [string], bits [int]
"mulaw": false, // apply mulaw if mode is bits
"padding": 2, // pad the input for resnet to see wider input length
// DATASET
//"use_gta": true, // use computed gta features from the tts model
"data_path": "tests/data/ljspeech/wavs/", // path containing training wav files
"feature_path": null, // path containing computed features from wav files if null compute them
// TRAINING
"batch_size": 4, // Batch size for training. Lower values than 32 might cause hard to learn attention.
"epochs": 1, // total number of epochs to train.
// VALIDATION
"run_eval": true,
"test_every_epochs": 10, // Test after set number of epochs (Test every 20 epochs for example)
// OPTIMIZER
"grad_clip": 4, // apply gradient clipping if > 0
"lr_scheduler": "MultiStepLR", // one of the schedulers from https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate
"lr_scheduler_params": {
"gamma": 0.5,
"milestones": [200000, 400000, 600000]
},
"lr": 1e-4, // initial learning rate
// TENSORBOARD and LOGGING
"print_step": 25, // Number of steps to log traning on console.
"print_eval": false, // If True, it prints loss values for each step in eval run.
"save_step": 25000, // Number of training steps expected to plot training stats on TB and save model checkpoints.
"checkpoint": true, // If true, it saves checkpoints per "save_step"
"tb_model_param_stats": false, // true, plots param stats per layer on tensorboard. Might be memory consuming, but good for debugging.
// DATA LOADING
"num_loader_workers": 4, // number of training data loader processes. Don't set it too big. 4-8 are good values.
"num_val_loader_workers": 4, // number of evaluation data loader processes.
"eval_split_size": 10, // number of samples for testing
// PATHS
"output_path": "tests/train_outputs/"
}

View File

@ -5,11 +5,11 @@ echo "$BASEDIR"
# create run dir
mkdir $BASEDIR/train_outputs
# run training
CUDA_VISIBLE_DEVICES="" python TTS/bin/train_vocoder.py --config_path $BASEDIR/inputs/test_vocoder_multiband_melgan_config.json
CUDA_VISIBLE_DEVICES="" python TTS/bin/train_gan_vocoder.py --config_path $BASEDIR/inputs/test_vocoder_multiband_melgan_config.json
# find the training folder
LATEST_FOLDER=$(ls $BASEDIR/train_outputs/| sort | tail -1)
echo $LATEST_FOLDER
# continue the previous training
CUDA_VISIBLE_DEVICES="" python TTS/bin/train_vocoder.py --continue_path $BASEDIR/train_outputs/$LATEST_FOLDER
CUDA_VISIBLE_DEVICES="" python TTS/bin/train_gan_vocoder.py --continue_path $BASEDIR/train_outputs/$LATEST_FOLDER
# remove all the outputs
rm -rf $BASEDIR/train_outputs/$LATEST_FOLDER

View File

@ -0,0 +1,31 @@
import numpy as np
import torch
import random
from TTS.vocoder.models.wavernn import WaveRNN
def test_wavernn():
model = WaveRNN(
rnn_dims=512,
fc_dims=512,
mode=10,
mulaw=False,
pad=2,
use_aux_net=True,
use_upsample_net=True,
upsample_factors=[4, 8, 8],
feat_dims=80,
compute_dims=128,
res_out_dims=128,
num_res_blocks=10,
hop_length=256,
sample_rate=22050,
)
dummy_x = torch.rand((2, 1280))
dummy_m = torch.rand((2, 80, 9))
y_size = random.randrange(20, 60)
dummy_y = torch.rand((80, y_size))
output = model(dummy_x, dummy_m)
assert np.all(output.shape == (2, 1280, 4 * 256)), output.shape
output = model.generate(dummy_y, True, 5500, 550, False)
assert np.all(output.shape == (256 * (y_size - 1),))

View File

@ -0,0 +1,92 @@
import os
import shutil
import numpy as np
from tests import get_tests_path, get_tests_input_path, get_tests_output_path
from torch.utils.data import DataLoader
from TTS.utils.audio import AudioProcessor
from TTS.utils.io import load_config
from TTS.vocoder.datasets.wavernn_dataset import WaveRNNDataset
from TTS.vocoder.datasets.preprocess import load_wav_feat_data, preprocess_wav_files
file_path = os.path.dirname(os.path.realpath(__file__))
OUTPATH = os.path.join(get_tests_output_path(), "loader_tests/")
os.makedirs(OUTPATH, exist_ok=True)
C = load_config(os.path.join(get_tests_input_path(),
"test_vocoder_wavernn_config.json"))
test_data_path = os.path.join(get_tests_path(), "data/ljspeech/")
test_mel_feat_path = os.path.join(test_data_path, "mel")
test_quant_feat_path = os.path.join(test_data_path, "quant")
ok_ljspeech = os.path.exists(test_data_path)
def wavernn_dataset_case(batch_size, seq_len, hop_len, pad, mode, mulaw, num_workers):
""" run dataloader with given parameters and check conditions """
ap = AudioProcessor(**C.audio)
C.batch_size = batch_size
C.mode = mode
C.seq_len = seq_len
C.data_path = test_data_path
preprocess_wav_files(test_data_path, C, ap)
_, train_items = load_wav_feat_data(
test_data_path, test_mel_feat_path, 5)
dataset = WaveRNNDataset(ap=ap,
items=train_items,
seq_len=seq_len,
hop_len=hop_len,
pad=pad,
mode=mode,
mulaw=mulaw
)
# sampler = DistributedSampler(dataset) if num_gpus > 1 else None
loader = DataLoader(dataset,
shuffle=True,
collate_fn=dataset.collate,
batch_size=batch_size,
num_workers=num_workers,
pin_memory=True,
)
max_iter = 10
count_iter = 0
try:
for data in loader:
x_input, mels, _ = data
expected_feat_shape = (ap.num_mels,
(x_input.shape[-1] // hop_len) + (pad * 2))
assert np.all(
mels.shape[1:] == expected_feat_shape), f" [!] {mels.shape} vs {expected_feat_shape}"
assert (mels.shape[2] - pad * 2) * hop_len == x_input.shape[1]
count_iter += 1
if count_iter == max_iter:
break
# except AssertionError:
# shutil.rmtree(test_mel_feat_path)
# shutil.rmtree(test_quant_feat_path)
finally:
shutil.rmtree(test_mel_feat_path)
shutil.rmtree(test_quant_feat_path)
def test_parametrized_wavernn_dataset():
''' test dataloader with different parameters '''
params = [
[16, C.audio['hop_length'] * 10, C.audio['hop_length'], 2, 10, True, 0],
[16, C.audio['hop_length'] * 10, C.audio['hop_length'], 2, "mold", False, 4],
[1, C.audio['hop_length'] * 10, C.audio['hop_length'], 2, 9, False, 0],
[1, C.audio['hop_length'], C.audio['hop_length'], 2, 10, True, 0],
[1, C.audio['hop_length'], C.audio['hop_length'], 2, "mold", False, 0],
[1, C.audio['hop_length'] * 5, C.audio['hop_length'], 4, 10, False, 2],
[1, C.audio['hop_length'] * 5, C.audio['hop_length'], 2, "mold", False, 0],
]
for param in params:
print(param)
wavernn_dataset_case(*param)

View File

@ -0,0 +1,15 @@
#!/usr/bin/env bash
BASEDIR=$(dirname "$0")
echo "$BASEDIR"
# create run dir
mkdir $BASEDIR/train_outputs
# run training
CUDA_VISIBLE_DEVICES="" python TTS/bin/train_wavernn_vocoder.py --config_path $BASEDIR/inputs/test_vocoder_wavernn_config.json
# find the training folder
LATEST_FOLDER=$(ls $BASEDIR/train_outputs/| sort | tail -1)
echo $LATEST_FOLDER
# continue the previous training
CUDA_VISIBLE_DEVICES="" python TTS/bin/train_wavernn_vocoder.py --continue_path $BASEDIR/train_outputs/$LATEST_FOLDER
# remove all the outputs
rm -rf $BASEDIR/train_outputs/$LATEST_FOLDER