fix: cut.load_audio needs to be an array

app.py

@@ -26,7 +26,8 @@ def transcribe_audio(audio_filepath):
         cut = cut.to_mono(mono_downmix=True)
 
     # Load audio data
-    audio, audio_lens = cut.load_audio()
+    audio = cut.load_audio()
+    audio_lens = audio.shape[0]
 
     # Generate transcription
     with torch.inference_mode():