add new app

app.py

@@ -0,0 +1,16 @@
+import gradio as gr
+from summarize import Summarizer
+
+interface = gr.Interface(fn = Summarizer,
+                        inputs = [gr.inputs.Textbox(lines=2, 
+                                                    placeholder="Enter your text...",
+                                                    label='YouTube Video Link'),
+                                  gr.inputs.Radio(["mT5", "BART", "Pegasus"], type="value", label='Model')],
+                        outputs = [gr.outputs.Textbox(
+                                                      label="Sar")],
+                         
+                        title = "Youtube Summarizer",
+                        examples = [['https://www.youtube.com/watch?v=A4OmtyaBHFE', 'mT5'],
+                                   ['https://www.youtube.com/watch?v=cU6xVZfkcgo', 'mT5']])
+
+interface.launch(debug=True) 

requirements.txt

@@ -0,0 +1,4 @@
+torch
+transformers
+sentencepiece
+youtube-transcript-api 

summarize.py

@@ -0,0 +1,36 @@
+from youtube_transcript_api import YouTubeTranscriptApi
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+
+def Summarizer(link, model):
+  
+  video_id = link.split("=")[1]
+
+  try:
+    transcript = YouTubeTranscriptApi.get_transcript(video_id)
+    FinalTranscript = ' '.join([i['text'] for i in transcript])
+    
+    if model == "Pegasus":
+      checkpoint = "google/pegasus-large"
+    elif model == "mT5":
+      checkpoint = "csebuetnlp/mT5_multilingual_XLSum"
+    elif model == "BART":
+      checkpoint = "sshleifer/distilbart-cnn-12-6"
+      
+    tokenizer = AutoTokenizer.from_pretrained(checkpoint)
+    model = AutoModelForSeq2SeqLM.from_pretrained(checkpoint)
+
+
+    inputs = tokenizer(FinalTranscript, 
+                    max_length=1024, 
+                    truncation=True,
+                    return_tensors="pt")
+    
+    summary_ids = model.generate(inputs["input_ids"])
+    summary = tokenizer.batch_decode(summary_ids, 
+                                  skip_special_tokens=True, 
+                                  clean_up_tokenization_spaces=False)
+    
+
+    return summary[0]
+  except Exception as e:
+    return "TranscriptsDisabled: Transcript is not available \nTry another video"