1  """
  2  Audio extraction module for video transcription
  3  -----------------------------------------------------
  4  This module provides functions for audio extraction from video files.
  5  """
  6  
  7  import os
  8  import logging
  9  import traceback
 10  from pathlib import Path
 11  from tempfile import NamedTemporaryFile
 12  from typing import Optional, Callable, Union
 13  from moviepy import AudioFileClip
 14  MOVIEPY_AVAILABLE = True
 15  # Logging configuration
 16  logger = logging.getLogger("transcription.audio_extraction")
 17  
 18  
 19  
 20  # Path configuration
 21  AUDIO_TMP_DIR = Path("uploads/audio")
 22  AUDIO_TMP_DIR.mkdir(parents=True, exist_ok=True)
 23  
 24  def extract_audio(
 25      video_path: str, 
 26      audio_path: Optional[str] = None, 
 27      progress: Optional[Callable] = None,
 28      audio_format: str = "wav",
 29      codec: str = "pcm_s16le"
 30  ) -> str:
 31      """
 32      Extracts audio from a video and saves it to a file
 33      
 34      Args:
 35          video_path: Path to the video file
 36          audio_path: Output path for the audio file (optional)
 37          progress: Progress tracking function (optional)
 38          audio_format: Audio file format (default: "wav")
 39          codec: Audio codec to use (default: "pcm_s16le")
 40          
 41      Returns:
 42          Path to the extracted audio file
 43          
 44      Raises:
 45          ImportError: If MoviePy is not available
 46          Exception: If an error occurs during extraction
 47      """
 48      if not MOVIEPY_AVAILABLE:
 49          raise ImportError("MoviePy is required for audio extraction")
 50      
 51      if progress:
 52          progress(0.1, desc="Extracting audio from video...")
 53      
 54      try:
 55          # Create a temporary file if no path is specified
 56          if audio_path is None:
 57              audio_file = NamedTemporaryFile(delete=False, suffix=f".{audio_format}", dir=AUDIO_TMP_DIR)
 58              audio_path = audio_file.name
 59              audio_file.close()
 60          
 61          # Extract audio
 62          audio = AudioFileClip(video_path)
 63          audio.write_audiofile(audio_path, codec=codec, verbose=False, logger=None)
 64          
 65          if progress:
 66              progress(0.3, desc="Audio extraction completed")
 67          
 68          return audio_path
 69          
 70      except Exception as e:
 71          error_msg = f"Error during audio extraction: {str(e)}"
 72          logger.error(error_msg)
 73          logger.error(traceback.format_exc())
 74          raise Exception(error_msg)
 75  
 76  def cleanup_audio_file(audio_path: str) -> bool:
 77      """
 78      Deletes a temporary audio file
 79      
 80      Args:
 81          audio_path: Path to the audio file to delete
 82          
 83      Returns:
 84          True if the file was deleted, False otherwise
 85      """
 86      if audio_path and isinstance(audio_path, str) and os.path.exists(audio_path):
 87          # Check if it's a temporary file in our directory
 88          if audio_path.startswith(str(AUDIO_TMP_DIR)):
 89              try:
 90                  os.unlink(audio_path)
 91                  return True
 92              except Exception as e:
 93                  logger.warning(f"Unable to delete temporary audio file: {str(e)}")
 94      return False
 95  
 96  def get_audio_duration(audio_path: str) -> float:
 97      """
 98      Gets the duration of an audio file in seconds
 99      
100      Args:
101          audio_path: Path to the audio file
102          
103      Returns:
104          Duration in seconds
105          
106      Raises:
107          ImportError: If MoviePy is not available
108      """
109      if not MOVIEPY_AVAILABLE:
110          raise ImportError("MoviePy is required to get audio duration")
111      
112      try:
113          audio = AudioFileClip(audio_path)
114          duration = audio.duration
115          audio.close()
116          return duration
117      except Exception as e:
118          logger.error(f"Error getting audio duration: {str(e)}")
119          raise