Update app.py

app.py

@@ -7,6 +7,7 @@ import tempfile
 import os
 import soundfile as sf
 from spellchecker import SpellChecker
+from pydub import AudioSegment
 
 # Check if CUDA is available and set the device
 device = "cuda" if torch.cuda.is_available() else "cpu"
@@ -87,9 +88,12 @@ def transcribe_video(url):
         audio_bytes = download_audio_from_url(url)
         print(f"Successfully downloaded {len(audio_bytes)} bytes of audio data")
         
-        # Save audio bytes to a temporary file
+        # Convert audio bytes to AudioSegment
+        audio = AudioSegment.from_file(io.BytesIO(audio_bytes))
+        
+        # Save as WAV file
         with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_audio:
-            temp_audio.write(audio_bytes)
+            audio.export(temp_audio.name, format="wav")
             temp_audio_path = temp_audio.name
 
         print("Starting audio transcription...")