Spaces:
Sleeping
Sleeping
File size: 2,710 Bytes
e613cea |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 |
from __future__ import annotations
import json
import os
from copy import deepcopy
from logging import getLogger
from pathlib import Path
import numpy as np
from librosa import get_duration
from tqdm import tqdm
LOG = getLogger(__name__)
CONFIG_TEMPLATE_DIR = Path(__file__).parent / "config_templates"
def preprocess_config(
input_dir: Path | str,
train_list_path: Path | str,
val_list_path: Path | str,
test_list_path: Path | str,
config_path: Path | str,
config_name: str,
):
input_dir = Path(input_dir)
train_list_path = Path(train_list_path)
val_list_path = Path(val_list_path)
test_list_path = Path(test_list_path)
config_path = Path(config_path)
train = []
val = []
test = []
spk_dict = {}
spk_id = 0
random = np.random.RandomState(1234)
for speaker in os.listdir(input_dir):
spk_dict[speaker] = spk_id
spk_id += 1
paths = []
for path in tqdm(list((input_dir / speaker).rglob("*.wav"))):
if get_duration(filename=path) < 0.3:
LOG.warning(f"skip {path} because it is too short.")
continue
paths.append(path)
random.shuffle(paths)
if len(paths) <= 4:
raise ValueError(
f"too few files in {input_dir / speaker} (expected at least 5)."
)
train += paths[2:-2]
val += paths[:2]
test += paths[-2:]
LOG.info(f"Writing {train_list_path}")
train_list_path.parent.mkdir(parents=True, exist_ok=True)
train_list_path.write_text(
"\n".join([x.as_posix() for x in train]), encoding="utf-8"
)
LOG.info(f"Writing {val_list_path}")
val_list_path.parent.mkdir(parents=True, exist_ok=True)
val_list_path.write_text("\n".join([x.as_posix() for x in val]), encoding="utf-8")
LOG.info(f"Writing {test_list_path}")
test_list_path.parent.mkdir(parents=True, exist_ok=True)
test_list_path.write_text("\n".join([x.as_posix() for x in test]), encoding="utf-8")
config = deepcopy(
json.loads(
(
CONFIG_TEMPLATE_DIR
/ (
config_name
if config_name.endswith(".json")
else config_name + ".json"
)
).read_text(encoding="utf-8")
)
)
config["spk"] = spk_dict
config["data"]["training_files"] = train_list_path.as_posix()
config["data"]["validation_files"] = val_list_path.as_posix()
LOG.info(f"Writing {config_path}")
config_path.parent.mkdir(parents=True, exist_ok=True)
with config_path.open("w", encoding="utf-8") as f:
json.dump(config, f, indent=2)
|