tuna2134/sbv2-v2
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

Feat: support various audio format in slice.py

Browse Source
This commit is contained in:
litagin02
2024-03-15 12:53:24 +09:00
parent 432edcc52b
commit b02b882c04
1 changed files with 12 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
slice.py
View File

@@ -14,6 +14,11 @@ from style_bert_vits2.logging import logger
from style_bert_vits2.utils.stdout_wrapper import SAFE_STDOUT from style_bert_vits2.utils.stdout_wrapper import SAFE_STDOUT
def is_audio_file(file: Path) -> bool:
supported_extensions = [".wav", ".flac", ".mp3", ".ogg", ".opus"]
return file.suffix.lower() in supported_extensions
def get_stamps( def get_stamps(
vad_model: Any, vad_model: Any,
utils: Any, utils: Any,
@@ -158,9 +163,9 @@ if __name__ == "__main__":
time_suffix: bool = args.time_suffix time_suffix: bool = args.time_suffix
num_processes: int = args.num_processes num_processes: int = args.num_processes
wav_files = Path(input_dir).glob("**/*.wav") audio_files = [file for file in input_dir.rglob("*") if is_audio_file(file)]
wav_files = list(wav_files)
logger.info(f"Found {len(wav_files)} wav files.") logger.info(f"Found {len(audio_files)} audio files.")
if output_dir.exists(): if output_dir.exists():
logger.warning(f"Output directory {output_dir} already exists, deleting...") logger.warning(f"Output directory {output_dir} already exists, deleting...")
shutil.rmtree(output_dir) shutil.rmtree(output_dir)
@@ -216,7 +221,7 @@ if __name__ == "__main__":
error_queue: Queue[tuple[Path, Exception]] = Queue() error_queue: Queue[tuple[Path, Exception]] = Queue()
# ファイル数が少ない場合は、ワーカー数をファイル数に合わせる # ファイル数が少ない場合は、ワーカー数をファイル数に合わせる
num_processes = min(num_processes, len(wav_files)) num_processes = min(num_processes, len(audio_files))
threads = [ threads = [
Thread(target=process_queue, args=(q, result_queue, error_queue)) Thread(target=process_queue, args=(q, result_queue, error_queue))
@@ -225,14 +230,14 @@ if __name__ == "__main__":
for t in threads: for t in threads:
t.start() t.start()
pbar = tqdm(total=len(wav_files), file=SAFE_STDOUT) pbar = tqdm(total=len(audio_files), file=SAFE_STDOUT)
for file in wav_files: for file in audio_files:
q.put(file) q.put(file)
# result_queueを監視し、要素が追加されるごとに結果を加算しプログレスバーを更新 # result_queueを監視し、要素が追加されるごとに結果を加算しプログレスバーを更新
total_sec = 0 total_sec = 0
total_count = 0 total_count = 0
for _ in range(len(wav_files)): for _ in range(len(audio_files)):
time, count = result_queue.get() time, count = result_queue.get()
total_sec += time total_sec += time
total_count += count total_count += count