Bare `except:` catches BaseException including KeyboardInterrupt and SystemExit. Replaced 9 instances with `except Exception:`.pull/4154/head
haosenwang1018
5 months ago
parent
d02ae35dc0
commit
ab6031e8d5
@ -1 +0,0 @@
|
||||
../../../other/tts_finetune/tts3/local/prepare_env.py
|
||||
@ -0,0 +1,62 @@
|
||||
# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
import argparse
|
||||
import os
|
||||
from pathlib import Path
|
||||
|
||||
|
||||
def generate_finetune_env(output_dir: Path, pretrained_model_dir: Path):
|
||||
|
||||
output_dir = output_dir / "checkpoints/"
|
||||
output_dir = output_dir.resolve()
|
||||
output_dir.mkdir(parents=True, exist_ok=True)
|
||||
|
||||
model_path = sorted(list((pretrained_model_dir).rglob("*.pdz")))[0]
|
||||
model_path = model_path.resolve()
|
||||
iter = int(str(model_path).split("_")[-1].split(".")[0])
|
||||
model_file = str(model_path).split("/")[-1]
|
||||
|
||||
os.system("cp %s %s" % (model_path, output_dir))
|
||||
|
||||
records_file = output_dir / "records.jsonl"
|
||||
with open(records_file, "w") as f:
|
||||
line = "\"time\": \"2022-08-06 07:51:53.463650\", \"path\": \"%s\", \"iteration\": %d" % (
|
||||
str(output_dir / model_file), iter)
|
||||
f.write("{" + line + "}" + "\n")
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
# parse config and args
|
||||
parser = argparse.ArgumentParser(
|
||||
description="Preprocess audio and then extract features.")
|
||||
|
||||
parser.add_argument(
|
||||
"--pretrained_model_dir",
|
||||
type=str,
|
||||
default="./pretrained_models/fastspeech2_aishell3_ckpt_1.1.0",
|
||||
help="Path to pretrained model")
|
||||
|
||||
parser.add_argument(
|
||||
"--output_dir",
|
||||
type=str,
|
||||
default="./exp/default/",
|
||||
help="directory to save finetune model.")
|
||||
|
||||
args = parser.parse_args()
|
||||
|
||||
output_dir = Path(args.output_dir).expanduser()
|
||||
output_dir.mkdir(parents=True, exist_ok=True)
|
||||
pretrained_model_dir = Path(args.pretrained_model_dir).expanduser()
|
||||
|
||||
generate_finetune_env(output_dir, pretrained_model_dir)
|
||||
@ -1 +0,0 @@
|
||||
../transformer_tts/normalize.py
|
||||
@ -0,0 +1,124 @@
|
||||
# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
"""Normalize feature files and dump them."""
|
||||
import argparse
|
||||
import logging
|
||||
from operator import itemgetter
|
||||
from pathlib import Path
|
||||
|
||||
import jsonlines
|
||||
import numpy as np
|
||||
from sklearn.preprocessing import StandardScaler
|
||||
from tqdm import tqdm
|
||||
|
||||
from paddlespeech.t2s.datasets.data_table import DataTable
|
||||
|
||||
|
||||
def main():
|
||||
"""Run preprocessing process."""
|
||||
parser = argparse.ArgumentParser(
|
||||
description="Normalize dumped raw features (See detail in parallel_wavegan/bin/normalize.py)."
|
||||
)
|
||||
parser.add_argument(
|
||||
"--metadata",
|
||||
type=str,
|
||||
required=True,
|
||||
help="directory including feature files to be normalized. "
|
||||
"you need to specify either *-scp or rootdir.")
|
||||
|
||||
parser.add_argument(
|
||||
"--dumpdir",
|
||||
type=str,
|
||||
required=True,
|
||||
help="directory to dump normalized feature files.")
|
||||
parser.add_argument(
|
||||
"--speech-stats",
|
||||
type=str,
|
||||
required=True,
|
||||
help="speech statistics file.")
|
||||
parser.add_argument(
|
||||
"--phones-dict", type=str, default=None, help="phone vocabulary file.")
|
||||
parser.add_argument(
|
||||
"--speaker-dict", type=str, default=None, help="speaker id map file.")
|
||||
|
||||
args = parser.parse_args()
|
||||
|
||||
# check directory existence
|
||||
dumpdir = Path(args.dumpdir).resolve()
|
||||
dumpdir.mkdir(parents=True, exist_ok=True)
|
||||
|
||||
# get dataset
|
||||
with jsonlines.open(args.metadata, 'r') as reader:
|
||||
metadata = list(reader)
|
||||
dataset = DataTable(
|
||||
metadata, converters={
|
||||
"speech": np.load,
|
||||
})
|
||||
logging.info(f"The number of files = {len(dataset)}.")
|
||||
|
||||
# restore scaler
|
||||
speech_scaler = StandardScaler()
|
||||
speech_scaler.mean_ = np.load(args.speech_stats)[0]
|
||||
speech_scaler.scale_ = np.load(args.speech_stats)[1]
|
||||
speech_scaler.n_features_in_ = speech_scaler.mean_.shape[0]
|
||||
|
||||
vocab_phones = {}
|
||||
with open(args.phones_dict, 'rt') as f:
|
||||
phn_id = [line.strip().split() for line in f.readlines()]
|
||||
for phn, id in phn_id:
|
||||
vocab_phones[phn] = int(id)
|
||||
|
||||
vocab_speaker = {}
|
||||
with open(args.speaker_dict, 'rt') as f:
|
||||
spk_id = [line.strip().split() for line in f.readlines()]
|
||||
for spk, id in spk_id:
|
||||
vocab_speaker[spk] = int(id)
|
||||
|
||||
# process each file
|
||||
output_metadata = []
|
||||
|
||||
for item in tqdm(dataset):
|
||||
utt_id = item['utt_id']
|
||||
speech = item['speech']
|
||||
# normalize
|
||||
speech = speech_scaler.transform(speech)
|
||||
speech_dir = dumpdir / "data_speech"
|
||||
speech_dir.mkdir(parents=True, exist_ok=True)
|
||||
speech_path = speech_dir / f"{utt_id}_speech.npy"
|
||||
np.save(speech_path, speech.astype(np.float32), allow_pickle=False)
|
||||
|
||||
phone_ids = [vocab_phones[p] for p in item['phones']]
|
||||
spk_id = vocab_speaker[item["speaker"]]
|
||||
record = {
|
||||
"utt_id": item['utt_id'],
|
||||
"spk_id": spk_id,
|
||||
"text": phone_ids,
|
||||
"text_lengths": item['text_lengths'],
|
||||
"speech_lengths": item['speech_lengths'],
|
||||
"speech": str(speech_path),
|
||||
}
|
||||
# add spk_emb for voice cloning
|
||||
if "spk_emb" in item:
|
||||
record["spk_emb"] = str(item["spk_emb"])
|
||||
output_metadata.append(record)
|
||||
output_metadata.sort(key=itemgetter('utt_id'))
|
||||
output_metadata_path = Path(args.dumpdir) / "metadata.jsonl"
|
||||
with jsonlines.open(output_metadata_path, 'w') as writer:
|
||||
for item in output_metadata:
|
||||
writer.write(item)
|
||||
logging.info(f"metadata dumped into {output_metadata_path}")
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
Loading…
Reference in new issue