Tried to add youtube video upload

app.py

@@ -1,9 +1,33 @@
 import spaces
 import gradio as gr
-# Use a pipeline as a high-level helper
+import os
+import logging
+from pytube import YouTube
 import torch
 from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline, AutoModelForCausalLM, AutoTokenizer
 
+def get_text(url):
+    if url != '':
+        output_text_transcribe = ''
+
+    yt = YouTube(url)
+    video = yt.streams.filter(only_audio=True).first()
+    out_file = video.download(output_path=".")
+
+    file_stats = os.stat(out_file)
+    logging.info(f'Size of audio file in Bytes: {file_stats.st_size}')
+    
+    if file_stats.st_size <= 30000000:
+        base, ext = os.path.splitext(out_file)
+        new_file = base + '.mp3'
+        os.rename(out_file, new_file)
+        a = new_file
+    
+        result = model.transcribe(a)
+        return result['text'].strip()
+    else:
+        logging.error('Videos for transcription on this space are limited to about 1.5 hours. Sorry about this limit but some joker thought they could stop this tool from working by transcribing many extremely long videos. Please visit https://steve.digital to contact me about this space.')
+
 @spaces.GPU(duration=60)
 def transcribe_audio(audio, model_id):
     if audio is None:
@@ -36,7 +60,6 @@ def transcribe_audio(audio, model_id):
     result = pipe(audio)
     return result["text"]
 
-# @spaces.GPU(duration=180)
 def proofread(text):
     if text is None:
         return "Please provide the transcribed text for proofreading."
@@ -50,15 +73,13 @@ def proofread(text):
     tokenizer = AutoTokenizer.from_pretrained("hfl/llama-3-chinese-8b-instruct-v3")
     model.to(device)
     
-    # Perform proofreading using the model
     input_text = prompt + text
     input_ids = tokenizer.encode(input_text, return_tensors="pt").to(device)
-    output = model.generate(input_ids, max_length=len(input_ids[0])+50, num_return_sequences=1, temperature=0.7)
+    output = model.generate(input_ids, max_length=len(input_ids[0]) + 50, num_return_sequences=1, temperature=0.7)
     proofread_text = tokenizer.decode(output[0], skip_special_tokens=True)
     
     return proofread_text
 
-
 with gr.Blocks() as demo:
     gr.Markdown("""
                 # Audio Transcription and Proofreading
@@ -70,7 +91,7 @@ with gr.Blocks() as demo:
     with gr.Row():
         with gr.Column():
             audio = gr.Audio(sources="upload", type="filepath")
-            video = gr.Video(sources="upload")
+            input_text_url = gr.Textbox(label="Video URL")
         model_dropdown = gr.Dropdown(choices=["openai/whisper-large-v3", "alvanlii/whisper-small-cantonese"], value="openai/whisper-large-v3")
 
     transcribe_button = gr.Button("Transcribe")