Medix-AI/coqui-tts
Medix-AI
/
coqui-tts
mirror of https://github.com/coqui-ai/TTS.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

add mk annealing (mk attn loss contribution)

This commit is contained in:
Eren Golge 2018-04-25 05:36:00 -07:00
parent e7350346bf
commit c55900b4ad
3 changed files with 54 additions and 40 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

55
config.json
View File

@ -1,33 +1,34 @@
{ {
"num_mels": 80, "num_mels": 80,
"num_freq": 1025, "num_freq": 1025,
"sample_rate": 22050, "sample_rate": 22050,
"frame_length_ms": 50, "frame_length_ms": 50,
"frame_shift_ms": 12.5, "frame_shift_ms": 12.5,
"preemphasis": 0.97, "preemphasis": 0.97,
"min_level_db": -100, "min_level_db": -100,
"ref_level_db": 20, "ref_level_db": 20,
"embedding_size": 256, "embedding_size": 256,
"text_cleaner": "english_cleaners", "text_cleaner": "english_cleaners",
"epochs": 500, "epochs": 500,
"lr": 0.002, "lr": 0.002,
"warmup_steps": 4000, "warmup_steps": 4000,
"batch_size": 32, "batch_size": 32,
"eval_batch_size":32, "eval_batch_size":32,
"r": 5, "r": 5,
"mk": 1,
"griffin_lim_iters": 60, "griffin_lim_iters": 60,
"power": 1.2, "power": 1.2,
"dataset": "LJSpeech", "dataset": "LJSpeech",
"meta_file_train": "metadata_train.csv", "meta_file_train": "metadata_train.csv",
"meta_file_val": "metadata_val.csv", "meta_file_val": "metadata_val.csv",
"data_path": "/data/shared/KeithIto/LJSpeech-1.0/", "data_path": "/data/shared/KeithIto/LJSpeech-1.0/",
"min_seq_len": 0, "min_seq_len": 0,
"num_loader_workers": 8, "num_loader_workers": 8,
"checkpoint": true, "checkpoint": true,
"save_step": 908, "save_step": 908,
"output_path": "/data/shared/erogol_models/" "output_path": "/data/shared/erogol_models/"
} }

21
train.py
View File

@ -19,7 +19,8 @@ from tensorboardX import SummaryWriter
from utils.generic_utils import (Progbar, remove_experiment_folder, from utils.generic_utils import (Progbar, remove_experiment_folder,
create_experiment_folder, save_checkpoint, create_experiment_folder, save_checkpoint,
save_best_model, load_config, lr_decay, save_best_model, load_config, lr_decay,
count_parameters, check_update, get_commit_hash) count_parameters, check_update, get_commit_hash,
create_attn_mask)
from utils.model import get_param_size from utils.model import get_param_size
from utils.visual import plot_alignment, plot_spectrogram from utils.visual import plot_alignment, plot_spectrogram
from models.tacotron import Tacotron from models.tacotron import Tacotron
@ -90,6 +91,9 @@ def train(model, criterion, data_loader, optimizer, epoch):
params_group['lr'] = current_lr params_group['lr'] = current_lr
optimizer.zero_grad() optimizer.zero_grad()
# setup mk
mk = mk_decay(c.mk, c.epochs, epoch)
# convert inputs to variables # convert inputs to variables
text_input_var = Variable(text_input) text_input_var = Variable(text_input)
@ -105,19 +109,10 @@ def train(model, criterion, data_loader, optimizer, epoch):
linear_spec_var = linear_spec_var.cuda() linear_spec_var = linear_spec_var.cuda()
# create attention mask # create attention mask
# TODO: vectorize
N = text_input_var.shape[1] N = text_input_var.shape[1]
T = mel_spec_var.shape[1] // c.r T = mel_spec_var.shape[1] // c.r
M = np.zeros([N, T]) M = create_attn_mask(N, T, g)
for t in range(T):
for n in range(N):
val = 20 * np.exp(-pow((n/N)-(t/T), 2.0)/0.05)
M[n, t] = val
e_x = np.exp(M - np.max(M))
M = e_x / e_x.sum(axis=0) # only difference
M = Variable(torch.FloatTensor(M).t()).cuda()
M = torch.stack([M]*32)
# forward pass # forward pass
mel_output, linear_output, alignments =\ mel_output, linear_output, alignments =\
model.forward(text_input_var, mel_spec_var) model.forward(text_input_var, mel_spec_var)
@ -129,7 +124,7 @@ def train(model, criterion, data_loader, optimizer, epoch):
linear_spec_var[:, :, :n_priority_freq], linear_spec_var[:, :, :n_priority_freq],
mel_lengths_var) mel_lengths_var)
attention_loss = criterion(alignments, M, mel_lengths_var) attention_loss = criterion(alignments, M, mel_lengths_var)
loss = mel_loss + linear_loss + attention_loss loss = mel_loss + linear_loss + mk * attention_loss
# backpass and check the grad norm # backpass and check the grad norm
loss.backward() loss.backward()

18
utils/generic_utils.py
View File

@ -131,6 +131,24 @@ def lr_decay(init_lr, global_step, warmup_steps):
return lr return lr
def create_attn_mask(N, T, g=0.05):
r'''creating attn mask for guided attention'''
M = np.zeros([N, T])
for t in range(T):
for n in range(N):
val = 20 * np.exp(-pow((n/N)-(t/T), 2.0)/g)
M[n, t] = val
e_x = np.exp(M - np.max(M))
M = e_x / e_x.sum(axis=0) # only difference
M = Variable(torch.FloatTensor(M).t()).cuda()
M = torch.stack([M]*32)
return M
def mk_decay(init_mk, max_epoch, n_epoch):
return init_mk * ((max_epoch - n_epoch) / max_epoch)
def count_parameters(model): def count_parameters(model):
r"""Count number of trainable parameters in a network""" r"""Count number of trainable parameters in a network"""
return sum(p.numel() for p in model.parameters() if p.requires_grad) return sum(p.numel() for p in model.parameters() if p.requires_grad)