fuck it, find out

app.py

@@ -25,7 +25,9 @@ embedding_model = PretrainedSpeakerEmbedding(
 )
 
 def transcribe(audio, num_speakers):
-  path = convert_to_wav(audio)
+  path, error = convert_to_wav(audio)
+  if error is not None:
+    return error
 
   duration = get_duration(path)
   if duration > 4 * 60 * 60:
@@ -45,9 +47,16 @@ def transcribe(audio, num_speakers):
 
 def convert_to_wav(path):
   if path[-3:] != 'wav':
-    subprocess.call(['ffmpeg', '-i', path, 'audio.wav', '-y'])
-    path = 'audio.wav'
-  return path
+    new_path = '.'.join(path.split('.')[:-1]) + '.wav'
+    try:
+      subprocess.call(['ffmpeg', '-i', path, '-c', 'copy', new_path, '-y'])
+    except:
+      try:
+        subprocess.call(['ffmpeg', '-i', path, new_path, '-y'])
+      except:
+        return path, 'Error: Could not convert file to .wav'
+    path = new_path
+  return path, None
 
 def get_duration(path):
   with contextlib.closing(wave.open(path,'r')) as f: