[update] add sampling rate

dataset/process.py

@@ -14,6 +14,7 @@ def __init__(
  self,
  lyric_path: str,
  audio_path: str,
+ sample_rate: int = None,
  export_path: str = None,
  clean: bool = False,
  ):
@@ -22,13 +23,15 @@ def __init__(
  Args:
  lyric_path (str): the path to the lyrics folder
  audio_path (str): the path to the audio folder
+ sample_rate (int, optional): the sample rate of the audio. Defaults to None.
  export_path (str, optional): the path to export data. Defaults to None.
  clean (bool, optional): remove all data in the export path. Defaults to False.
  """
 
  self.lyric_path = lyric_path
  self.audio_path = audio_path
  self.export_path = export_path
+ self.sample_rate = sample_rate
 
  if clean:
  self.remove_export_folder()
@@ -123,6 +126,9 @@ def _export_audio(self, audios: List[AudioSegment], file_name: str) -> None:
  if self.export_path:
  path = f"{self.export_path}/audio/{file_name}_{i}.wav"
 
+ if self.sample_rate:
+ audio = audio.set_frame_rate(self.sample_rate)
+
  audio.export(path, format="wav")
 
  def _export_lyric(self, lyrics: List[str], file_name: str) -> None:

process.py → process_dataset.py

@@ -9,12 +9,14 @@
 parser.add_argument("--audio_path", type=str, default="dataset/audio")
 parser.add_argument("--lyric_path", type=str, default="dataset/lyrics")
 parser.add_argument("--export_path", type=str, default="dataset/export")
+parser.add_argument("--sample_rate", type=int, default=None)
 parser.add_argument("--clean", type=bool, default=False)
 
 args = parser.parse_args()
 process = DatasetProcess(
  lyric_path=args.lyric_path,
  audio_path=args.audio_path,
+ sample_rate=args.sample_rate,
  export_path=args.export_path,
  clean=args.clean,
 )