Update app.py

app.py

@@ -3,13 +3,7 @@ import streamlit.components.v1 as components
 import requests
 import os
 import time
-import torch
-from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
-from transformers import pipeline
 import streamlit as st
-import pytube as pt
-from transformers import AutoModelForSeq2SeqLM
-
 from streamlit_mic_recorder import mic_recorder
 MODEL_NAME = "drinktoomuchsax/whisper-small-hi"
 lang = "en"
@@ -27,50 +21,6 @@ BASETEN_KEY = os.environ.get("BASETEN_KEY", None)
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
 
-### Whisper Start
-
-pipe = pipeline(
-    task="automatic-speech-recognition",
-    model=MODEL_NAME,
-    chunk_length_s=30,
-    device=device,
-)
-pipe.model.config.forced_decoder_ids = pipe.tokenizer.get_decoder_prompt_ids(language=lang, task="transcribe")
-
-# Transcription function
-def transcribe(file_path):
-    # Load the audio file
-    inputs = tokenizer(file_path, return_tensors="pt", padding="longest", truncation=True)
-    inputs = {key: value.to(device) for key, value in inputs.items()}
-    
-    # Pass the inputs and the attention_mask to the model
-    generated_ids = pipe.model.generate(inputs["input_ids"], attention_mask=inputs["attention_mask"], max_length=1000)
-    transcription = tokenizer.decode(generated_ids[0], skip_special_tokens=True)
-    
-    return transcription
-
-# YouTube transcription function
-def yt_transcribe(yt_url):
-    yt = pt.YouTube(yt_url)
-    stream = yt.streams.filter(only_audio=True)[0]
-    stream.download(filename="audio.mp3")
-    transcription = transcribe("audio.mp3")
-    return transcription
-
-### Whisper END
-
-
-# Load the tokenizer and model
-tokenizer = AutoTokenizer.from_pretrained("merve/chatgpt-prompts-bart-long")
-model = AutoModelForSeq2SeqLM.from_pretrained("merve/chatgpt-prompts-bart-long", from_tf=True).to("cuda" if torch.cuda.is_available() else "cpu")
-
-# Function to generate the prompt based on the persona
-def generate(prompt):
-    batch = tokenizer(prompt, return_tensors="pt").to(model.device)
-    generated_ids = model.generate(batch["input_ids"], max_new_tokens=150)
-    output = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
-    return output[0]
-
 
 #st.set_page_config(layout="wide")
 # Load custom CSS to integrate Bootstrap, Font Awesome, and Google Fonts
@@ -157,13 +107,11 @@ with lr:
     if st.button("Generate Prompt"):
         if persona:
             with st.spinner("Generating..."):
-                result = generate(persona)
+                result = "Test"
             st.text_area("Generated Prompt", value=result, height=200)
         else:
             st.error("Please enter a persona to generate a prompt.")
 
-    
-
 with rl:
     # End of Box 2 and second Carousel Item
     st.markdown('''<h3><i class="fa fa-pencil"></i> Transcribe </h3>''', unsafe_allow_html=True)
@@ -186,11 +134,8 @@ with rl:
             with open("temp_recording.wav", "wb") as f:
                 f.write(audio["bytes"])
             with st.spinner("Transcribing..."):
-                transcription = transcribe("temp_recording.wav")
-            st.text_area("Transcription", transcription, height=200)
-        elif uploaded_file is not None:
-            with st.spinner("Transcribing..."):
-                transcription = transcribe(uploaded_file)
+                #transcription = transcribe("temp_recording.wav")
+                #need to send the data here
             st.text_area("Transcription", transcription, height=200)
         else:
             st.error("Please record audio or upload a file to transcribe.")