Update app.py

app.py

@@ -1,4 +1,5 @@
 import torch
+from pydub import AudioSegment
 from transformers import AutoModelForSequenceClassification, AutoTokenizer
 from transformers.pipelines.audio_utils import ffmpeg_read
 import gradio as gr
@@ -73,16 +74,33 @@ def transcribe(Microphone, File_Upload, URL):
     elif Microphone is not None:
         file = Microphone
         
+    #elif URL:
+    #    link = YouTube(URL)
+    #    file = link.streams.filter(only_audio=True)[0].download(filename="audio.mp3")
     elif URL:
         link = YouTube(URL)
-        file = link.streams.filter(only_audio=True)[0].download(filename="audio.mp4")
+        stream = link.streams.filter(only_audio=True).first()
+        
+        # Download the audio file with a temporary filename
+        temp_filename = "temp_audio_file"
+        stream.download(filename=temp_filename)
+        
+        # Load the downloaded file with pydub and convert it to mp3
+        audio = AudioSegment.from_file(temp_filename, format="mp4")
+        
+        # Truncate it to the first 30 seconds
+        truncated_audio = audio[:30000]  # AudioSegment works in milliseconds
+        
+        file = "file.mp3"
+        truncated_audio.export(file, format="mp3")   
+
     else:
         file = File_Upload
     
 
     language = None
 
-    options = whisper.DecodingOptions(without_timestamps=True)
+    options = whisper.DecodingOptions()
 
     loaded_model = whisper.load_model("base")
     transcript = loaded_model.transcribe(file, language=language)