Update app.py

app.py

@@ -3,26 +3,25 @@
 import streamlit as st
 import whisper
 from pytube import YouTube
-# from transformers import pipeline
 import os
 import openai
 import pinecone
 from sentence_transformers import SentenceTransformer
 
-# get API key from top-right dropdown on OpenAI website
+# openAI setup
 openai.api_key = "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"
 
-# pinecode
+# pinecode setup
 pinecone.init(api_key="df54ffd1-2104-410e-b210-44f12f0c2dfd", environment="us-west1-gcp")
 index = pinecone.Index("question-search")
 
-
-# load model
+# load model for transcription
 model = whisper.load_model("base")
-# qa_model = pipeline("question-answering")
-model_tf = SentenceTransformer('all-mpnet-base-v2')
 
+# load model for embedding
+model_tf = SentenceTransformer('all-mpnet-base-v2')
 
+# function for complete prompt
 def complete(prompt):
     # query text-davinci-003
     res = openai.Completion.create(
@@ -37,6 +36,7 @@ def complete(prompt):
     )
     return res['choices'][0]['text'].strip()
 
+# function to build context
 def query_context(prompt):  
   embeding_prompt = model_tf.encode(prompt)
   embeding_prompt = embeding_prompt.tolist()
@@ -51,18 +51,19 @@ def query_context(prompt):
   query_with_contexts =  f"Answer the question based on the context below, and if the question can't be answered based on the context, say \"I don't know\"\n\nContext: {resp['matches'][0]['metadata']['sentence']}\n\n---\n\nQuestion: {prompt}\nAnswer:"
   return query_with_contexts
 
-
-
-st.markdown("Upload Youtube URL")
-
-# upload voice clip
+# take video 
+st.markdown("Upload Youtube URL.........")
 audio_file = st.text_input("Upload an audio file")
+
 # write a question
+st.markdown("Write Query ........")
 prompt = st.text_input("Wtite a question...")
 
 # if button is submit
 if st.button('Submit', key=1):
 
+    ###################### TRANSCRIPTION ############################
+    
     # grab the youtube video and name
     youtube_video = YouTube(audio_file)
     name = youtube_video.title
@@ -82,13 +83,10 @@ if st.button('Submit', key=1):
     st.markdown("Transcription")
     st.write(transcription)
 
-    
-    context = transcription
-    # st.write("Answer:")
-    # answer = qa_model(question = question, context = context)['answer']
-    # st.write(answer)
+    ###################### QA ############################
 
-    # split document into sentences
+    # tokenize the context
+    context = transcription
     sentences = context.split(".")
 
     # loop through sentences
@@ -104,9 +102,10 @@ if st.button('Submit', key=1):
         i = i+1
     
 
+    # build query with context
     full_query = query_context(prompt)
-
+    # pass centexed query
     output = complete(full_query)
-
+    # output answer
     st.markdown("Answer")
     st.write(output)