Upload app.py

app.py

@@ -1,27 +1,28 @@
 import os
 import uuid
+import json
 import logging
-import streamlit as st
 import requests
+import streamlit as st
 from dotenv import load_dotenv
 from utils import voice_map, get_voice_prompt_style, AUDIO_DIR
 from generate_audio import generate_audio
 
-# Load secrets
+# Load API keys from .env
 load_dotenv()
 OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
 ELEVENLABS_API_KEY = os.getenv("ELEVENLABS_API_KEY")
 
-# Setup
+# Ensure audio output folder exists
 os.makedirs(AUDIO_DIR, exist_ok=True)
 logging.basicConfig(filename="frontend.log", level=logging.INFO)
 
-# Page config
+# Streamlit UI config
 st.set_page_config(page_title="Voice Agent Pro", page_icon="🎤")
 st.title("🎙️ Voice Agent Pro")
 st.markdown("Summarized answers with expressive AI voices.")
 
-# Sidebar voice selector
+# Sidebar: select voice
 st.sidebar.header("🎚️ Voice Settings")
 voice_label = st.sidebar.selectbox("Choose a voice:", list(voice_map.keys()))
 voice_id = voice_map[voice_label]
@@ -33,18 +34,12 @@ if "audio_key" not in st.session_state: st.session_state.audio_key = None
 if "file_text" not in st.session_state: st.session_state.file_text = ""
 if "key_points" not in st.session_state: st.session_state.key_points = []
 
-# Query box
-query = st.text_area(
-    "🗨️ Ask or refine something based on the bullets:",
-    value="",
-    placeholder="e.g., What makes you so cool, Grandma?",
-    key="query"
-)
-
+# Input fields
+query = st.text_area("🗨️ Ask or refine something based on the bullets:", value="", placeholder="e.g., What makes you so cool, Grandma?", key="query")
 url = st.text_input("🌐 Optional URL to summarize:")
 uploaded_file = st.file_uploader("📎 Or upload a file (PDF, TXT, DOCX)", type=["pdf", "txt", "docx"])
 
-# Clear all
+# Reset app state
 if st.button("🧹 Clear All"):
     st.session_state.query = ""
     st.session_state.file_text = ""
@@ -52,17 +47,25 @@ if st.button("🧹 Clear All"):
     st.session_state.audio_key = None
     st.session_state.key_points = []
 
-# Summarize
+# Helper: OpenAI response streaming
+def stream_openai_response(payload, headers):
+    with requests.post("https://api.openai.com/v1/chat/completions", headers=headers, json=payload, stream=True) as r:
+        for line in r.iter_lines():
+            if line and line.startswith(b"data: "):
+                yield line[len(b"data: "):].decode()
+
+# Main button: summarize and speak
 if st.button("🔁 Summarize"):
     if not query and not url and not uploaded_file:
         st.warning("Please enter a question, a URL, or upload a file.")
     else:
         with st.spinner("Talking to GPT..."):
             try:
+                # Load file content if present
                 if uploaded_file:
                     st.session_state.file_text = uploaded_file.read().decode("utf-8")
 
-                # Compose prompt
+                # Build GPT prompt
                 context = ""
                 if st.session_state.file_text:
                     context += st.session_state.file_text + "\n\n"
@@ -70,29 +73,34 @@ if st.button("🔁 Summarize"):
                     context += f"Summarize this page: {url}\n\n"
                 context += f"{tone_prompt}\n\nNow answer: {query}"
 
-                # GPT call
+                # OpenAI request setup
                 headers = {"Authorization": f"Bearer {OPENAI_API_KEY}"}
-                response = requests.post(
-                    "https://api.openai.com/v1/chat/completions",
-                    headers=headers,
-                    json={
-                        "model": "gpt-4o",
-                        "messages": [{"role": "user", "content": context}],
-                        "temperature": 0.7
-                    }
-                )
-                answer = response.json()["choices"][0]["message"]["content"]
-                st.session_state.answer = answer
-
-                # Generate audio
+                payload = {
+                    "model": "gpt-4o",
+                    "messages": [{"role": "user", "content": context}],
+                    "temperature": 0.7,
+                    "stream": True
+                }
+
+                # Streaming UI update
+                st.session_state.answer = ""
+                answer_box = st.empty()
+
+                for chunk in stream_openai_response(payload, headers):
+                    parsed = json.loads(chunk)
+                    delta = parsed['choices'][0]['delta'].get('content', '')
+                    st.session_state.answer += delta
+                    answer_box.markdown(st.session_state.answer)
+
+                # Generate audio from final answer
                 audio_key = str(uuid.uuid4())
-                generate_audio(answer, voice_id, audio_key)
+                generate_audio(st.session_state.answer, voice_id, audio_key)
                 st.session_state.audio_key = audio_key
 
             except Exception as e:
                 st.error(f"🔥 Error: {e}")
 
-# Display answer
+# Final UI display
 if st.session_state.answer:
     st.subheader("📜 Answer")
     st.success(st.session_state.answer)