Lint bert gen & remove unused vars
This commit is contained in:
25
bert_gen.py
25
bert_gen.py
@@ -1,23 +1,21 @@
|
|||||||
import torch
|
import torch
|
||||||
from torch.utils.data import DataLoader
|
|
||||||
from multiprocessing import Pool
|
from multiprocessing import Pool
|
||||||
import commons
|
import commons
|
||||||
import utils
|
import utils
|
||||||
from data_utils import TextAudioSpeakerLoader, TextAudioSpeakerCollate
|
|
||||||
from tqdm import tqdm
|
from tqdm import tqdm
|
||||||
import warnings
|
|
||||||
|
|
||||||
from text import cleaned_text_to_sequence, get_bert
|
from text import cleaned_text_to_sequence, get_bert
|
||||||
|
|
||||||
config_path = 'configs/config.json'
|
config_path = "configs/config.json"
|
||||||
hps = utils.get_hparams_from_file(config_path)
|
hps = utils.get_hparams_from_file(config_path)
|
||||||
|
total_gpus = torch.cuda.device_count()
|
||||||
|
|
||||||
|
|
||||||
def process_line(line):
|
def process_line(line):
|
||||||
_id, spk, language_str, text, phones, tone, word2ph = line.strip().split("|")
|
wav_path, _, language_str, text, phones, tone, word2ph = line.strip().split("|")
|
||||||
phone = phones.split(" ")
|
phone = phones.split(" ")
|
||||||
tone = [int(i) for i in tone.split(" ")]
|
tone = [int(i) for i in tone.split(" ")]
|
||||||
word2ph = [int(i) for i in word2ph.split(" ")]
|
word2ph = [int(i) for i in word2ph.split(" ")]
|
||||||
w2pho = [i for i in word2ph]
|
|
||||||
word2ph = [i for i in word2ph]
|
word2ph = [i for i in word2ph]
|
||||||
phone, tone, language = cleaned_text_to_sequence(phone, tone, language_str)
|
phone, tone, language = cleaned_text_to_sequence(phone, tone, language_str)
|
||||||
|
|
||||||
@@ -28,26 +26,27 @@ def process_line(line):
|
|||||||
for i in range(len(word2ph)):
|
for i in range(len(word2ph)):
|
||||||
word2ph[i] = word2ph[i] * 2
|
word2ph[i] = word2ph[i] * 2
|
||||||
word2ph[0] += 1
|
word2ph[0] += 1
|
||||||
wav_path = f'{_id}'
|
|
||||||
|
|
||||||
bert_path = wav_path.replace(".wav", ".bert.pt")
|
bert_path = wav_path.replace(".wav", ".bert.pt")
|
||||||
|
|
||||||
try:
|
try:
|
||||||
bert = torch.load(bert_path)
|
bert = torch.load(bert_path)
|
||||||
assert bert.shape[-1] == len(phone)
|
assert bert.shape[-1] == len(phone)
|
||||||
except:
|
except Exception:
|
||||||
bert = get_bert(text, word2ph, language_str)
|
bert = get_bert(text, word2ph, language_str)
|
||||||
assert bert.shape[-1] == len(phone)
|
assert bert.shape[-1] == len(phone)
|
||||||
torch.save(bert, bert_path)
|
torch.save(bert, bert_path)
|
||||||
|
|
||||||
|
|
||||||
if __name__ == '__main__':
|
if __name__ == "__main__":
|
||||||
lines = []
|
lines = []
|
||||||
with open(hps.data.training_files, encoding='utf-8' ) as f:
|
with open(hps.data.training_files, encoding="utf-8") as f:
|
||||||
lines.extend(f.readlines())
|
lines.extend(f.readlines())
|
||||||
|
|
||||||
with open(hps.data.validation_files, encoding='utf-8' ) as f:
|
with open(hps.data.validation_files, encoding="utf-8") as f:
|
||||||
lines.extend(f.readlines())
|
lines.extend(f.readlines())
|
||||||
|
|
||||||
with Pool(processes=12) as pool: #A100 40GB suitable config,if coom,please decrease the processess number.
|
# A100 40GB suitable config, if coom, please decrease the processes number.
|
||||||
for _ in tqdm(pool.imap_unordered(process_line, lines)):
|
with Pool(processes=6) as pool:
|
||||||
|
for _ in tqdm(pool.imap_unordered(process_line, lines), total=len(lines)):
|
||||||
pass
|
pass
|
||||||
|
|||||||
Reference in New Issue
Block a user