Fix audio validation with video file

modules/utils/audio_manager.py

@@ -2,13 +2,27 @@
 import soundfile as sf
 import os
 import numpy as np
+from faster_whisper.audio import decode_audio
+
+from modules.utils.files_manager import is_video
+from modules.utils.logger import get_logger
+
+logger = get_logger()
 
 
 def validate_audio(audio: Optional[str] = None):
     """Validate audio file and check if it's corrupted"""
     if isinstance(audio, np.ndarray):
         return True
 
+    if is_video(audio):
+        try:
+            audio = decode_audio(audio)
+            return True
+        except Exception as e:
+            logger.info(f"The file {audio} is not able to open or corrupted. Please check the file. {e}")
+            return False
+
     if not os.path.exists(audio):
         return False
 
@@ -19,4 +33,5 @@ def validate_audio(audio: Optional[str] = None):
             else:
                 return False
     except Exception as e:
+        logger.info(f"The file {audio} is not able to open or corrupted. Please check the file. {e}")
         return False

modules/whisper/base_transcription_pipeline.py

@@ -113,7 +113,6 @@ def run(self,
         start_time = time.time()
 
         if not validate_audio(audio):
-            logger.info(f"The audio file {audio} is not able to open or corrupted. Please check the file.")
             return [Segment()], 0
 
         params = TranscriptionPipelineParams.from_list(list(pipeline_params))