-
Notifications
You must be signed in to change notification settings - Fork 4
/
xtts_generate_dataset.py
60 lines (49 loc) · 2.33 KB
/
xtts_generate_dataset.py
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
import os
import json
import argparse
import traceback
from scripts.utils.formatter import format_audio_list
def preprocess_audio_dataset(audio_dir, target_language, whisper_version, out_path, name):
# Create a subdirectory within out_path using the 'name' parameter
named_out_path = os.path.join(out_path, name)
# Ensure the named output path exists
os.makedirs(named_out_path, exist_ok=True)
# List audio files in the directory
audio_paths = [os.path.join(audio_dir, f) for f in os.listdir(audio_dir) if f.lower().endswith(('.wav', '.mp3', '.flac'))]
# Process the audio files
try:
train_meta, eval_meta, audio_total_size = format_audio_list(
audio_paths,
whisper_model=whisper_version,
target_language=target_language,
out_path=named_out_path,
speaker_name=name
)
print("ok3")
# Check if total audio length is sufficient
if audio_total_size < 120:
return "The sum of the duration of the audios should be at least 2 minutes!"
print("Dataset Processed!")
return "Dataset Processed!", train_meta, eval_meta
except Exception as e:
traceback.print_exc()
return f"The data processing was interrupted due to an error: {e}"
def load_dataset_config(json_file_path):
with open(json_file_path, 'r') as file:
return json.load(file)
if __name__ == "__main__":
parser = argparse.ArgumentParser(description="Generate datasets from audio files.")
parser.add_argument("--config", type=str, required=True, help="Path to the dataset configuration JSON file.")
parser.add_argument("--whisper_version", type=str, required=True, help="Whisper model version can be .")
args = parser.parse_args()
dataset_configs = load_dataset_config(args.config)
print(dataset_configs)
# Set the base output path to 'output_datasets' in the current directory
base_out_path = os.path.join(os.getcwd(), "output_datasets")
os.makedirs(base_out_path, exist_ok=True)
for config in dataset_configs:
if config.get("activate", True):
audio_dir = config["audio_path"]
language = config["language"]
name = config.get("name", "default_name")
preprocess_audio_dataset(audio_dir, language, args.whisper_version, base_out_path, name)