Update app.py

app.py

@@ -1,5 +1,19 @@
 import streamlit as st
-import anthropic, openai, base64, cv2, glob, json, math, os, pytz, random, re, requests, textract, time, zipfile
+import anthropic
+import openai
+import base64
+import cv2
+import glob
+import json
+import math
+import os
+import pytz
+import random
+import re
+import requests
+import textract
+import time
+import zipfile
 import plotly.graph_objects as go
 import streamlit.components.v1 as components
 from datetime import datetime
@@ -20,10 +34,13 @@ from streamlit.runtime.scriptrunner import get_script_run_ctx
 import asyncio
 import edge_tts
 from streamlit_marquee import streamlit_marquee
+from typing import Tuple, Optional
+import pandas as pd
 
 # ─────────────────────────────────────────────────────────
 # 1. CORE CONFIGURATION & SETUP
 # ─────────────────────────────────────────────────────────
+
 st.set_page_config(
     page_title="🚲TalkingAIResearcher🏆",
     page_icon="🚲🏆",
@@ -37,7 +54,7 @@ st.set_page_config(
 )
 load_dotenv()
 
-# Available English voices for Edge TTS
+# ▶ Available English voices for Edge TTS
 EDGE_TTS_VOICES = [
     "en-US-AriaNeural",
     "en-US-GuyNeural",
@@ -50,7 +67,7 @@ EDGE_TTS_VOICES = [
     "en-CA-LiamNeural"
 ]
 
-# Session state variables
+# ▶ Initialize Session State
 if 'marquee_settings' not in st.session_state:
     st.session_state['marquee_settings'] = {
         "background": "#1E1E1E",
@@ -60,53 +77,50 @@ if 'marquee_settings' not in st.session_state:
         "width": "100%",
         "lineHeight": "35px"
     }
-
 if 'tts_voice' not in st.session_state:
     st.session_state['tts_voice'] = EDGE_TTS_VOICES[0]
-
 if 'audio_format' not in st.session_state:
     st.session_state['audio_format'] = 'mp3'
-
 if 'transcript_history' not in st.session_state:
     st.session_state['transcript_history'] = []
-
 if 'chat_history' not in st.session_state:
     st.session_state['chat_history'] = []
-
 if 'openai_model' not in st.session_state:
     st.session_state['openai_model'] = "gpt-4o-2024-05-13"
-
 if 'messages' not in st.session_state:
     st.session_state['messages'] = []
-
 if 'last_voice_input' not in st.session_state:
     st.session_state['last_voice_input'] = ""
-
 if 'editing_file' not in st.session_state:
     st.session_state['editing_file'] = None
-
 if 'edit_new_name' not in st.session_state:
     st.session_state['edit_new_name'] = ""
-
 if 'edit_new_content' not in st.session_state:
     st.session_state['edit_new_content'] = ""
-
 if 'viewing_prefix' not in st.session_state:
     st.session_state['viewing_prefix'] = None
-
 if 'should_rerun' not in st.session_state:
     st.session_state['should_rerun'] = False
-
 if 'old_val' not in st.session_state:
     st.session_state['old_val'] = None
-
 if 'last_query' not in st.session_state:
     st.session_state['last_query'] = ""
-
 if 'marquee_content' not in st.session_state:
     st.session_state['marquee_content'] = "🚀 Welcome to TalkingAIResearcher | 🤖 Your Research Assistant"
 
-# API Keys
+# ▶ Additional keys for performance, caching, etc.
+if 'audio_cache' not in st.session_state:
+    st.session_state['audio_cache'] = {}
+if 'download_link_cache' not in st.session_state:
+    st.session_state['download_link_cache'] = {}
+if 'operation_timings' not in st.session_state:
+    st.session_state['operation_timings'] = {}
+if 'performance_metrics' not in st.session_state:
+    st.session_state['performance_metrics'] = defaultdict(list)
+if 'enable_audio' not in st.session_state:
+    st.session_state['enable_audio'] = True  # Turn TTS on/off
+
+# ▶ API Keys
 openai_api_key = os.getenv('OPENAI_API_KEY', "")
 anthropic_key = os.getenv('ANTHROPIC_API_KEY_3', "")
 xai_key = os.getenv('xai',"")
@@ -120,7 +134,7 @@ openai_client = OpenAI(api_key=openai.api_key, organization=os.getenv('OPENAI_OR
 HF_KEY = os.getenv('HF_KEY')
 API_URL = os.getenv('API_URL')
 
-# Helper constants
+# ▶ Helper constants
 FILE_EMOJIS = {
     "md": "📝",
     "mp3": "🎵",
@@ -128,20 +142,77 @@ FILE_EMOJIS = {
 }
 
 # ─────────────────────────────────────────────────────────
-# 2. HELPER FUNCTIONS
+# 2. PERFORMANCE MONITORING & TIMING
+# ─────────────────────────────────────────────────────────
+
+class PerformanceTimer:
+    """
+    ⏱️ A context manager for timing operations with automatic logging.
+    Usage:
+        with PerformanceTimer("my_operation"):
+            # do something
+    The duration is stored into `st.session_state['operation_timings']`
+    and appended to the `performance_metrics` list.
+    """
+    def __init__(self, operation_name: str):
+        self.operation_name = operation_name
+        self.start_time = None
+        
+    def __enter__(self):
+        self.start_time = time.time()
+        return self
+        
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        if not exc_type:  # Only log if no exception occurred
+            duration = time.time() - self.start_time
+            st.session_state['operation_timings'][self.operation_name] = duration
+            st.session_state['performance_metrics'][self.operation_name].append(duration)
+
+def log_performance_metrics():
+    """
+    📈 Display performance metrics in the sidebar, including a timing breakdown
+    and a small bar chart of average times.
+    """
+    st.sidebar.markdown("### ⏱️ Performance Metrics")
+    
+    metrics = st.session_state['operation_timings']
+    if metrics:
+        total_time = sum(metrics.values())
+        st.sidebar.write(f"**Total Processing Time:** {total_time:.2f}s")
+        
+        # Break down each operation time
+        for operation, duration in metrics.items():
+            percentage = (duration / total_time) * 100
+            st.sidebar.write(f"**{operation}:** {duration:.2f}s ({percentage:.1f}%)")
+            
+        # Show timing history chart
+        history_data = []
+        for op, times in st.session_state['performance_metrics'].items():
+            if times:  # Only if we have data
+                avg_time = sum(times) / len(times)
+                history_data.append({"Operation": op, "Avg Time (s)": avg_time})
+        
+        if history_data:
+            st.sidebar.markdown("### 📊 Timing History (Avg)")
+            chart_data = pd.DataFrame(history_data)
+            st.sidebar.bar_chart(chart_data.set_index("Operation"))
+
+# ─────────────────────────────────────────────────────────
+# 3. HELPER FUNCTIONS (FILENAMES, LINKS, MARQUEE, ETC.)
 # ─────────────────────────────────────────────────────────
 
 def get_central_time():
-    """Get current time in US Central timezone."""
+    """🌎 Get current time in US Central timezone."""
     central = pytz.timezone('US/Central')
     return datetime.now(central)
 
 def format_timestamp_prefix():
-    """Generate timestamp prefix in format MM_dd_yy_hh_mm_AM/PM."""
+    """📅 Generate a timestamp prefix: MM_dd_yy_hh_mm_AM/PM."""
     ct = get_central_time()
     return ct.strftime("%m_%d_%y_%I_%M_%p")
 
 def initialize_marquee_settings():
+    """🌈 Initialize marquee defaults if needed."""
     if 'marquee_settings' not in st.session_state:
         st.session_state['marquee_settings'] = {
             "background": "#1E1E1E",
@@ -153,11 +224,12 @@ def initialize_marquee_settings():
         }
 
 def get_marquee_settings():
+    """🔧 Retrieve marquee settings from session."""
     initialize_marquee_settings()
     return st.session_state['marquee_settings']
 
 def update_marquee_settings_ui():
-    """Add color pickers & sliders for marquee config in sidebar."""
+    """🖌 Add color pickers & sliders for marquee config in the sidebar."""
     st.sidebar.markdown("### 🎯 Marquee Settings")
     cols = st.sidebar.columns(2)
     with cols[0]:
@@ -169,7 +241,7 @@ def update_marquee_settings_ui():
                                     key="text_color_picker")
     with cols[1]:
         font_size = st.slider("📏 Size", 10, 24, 14, key="font_size_slider")
-        duration = st.slider("⏱️ Speed", 1, 20, 20, key="duration_slider")
+        duration = st.slider("⏱️ Speed (secs)", 1, 20, 20, key="duration_slider")
 
     st.session_state['marquee_settings'].update({
         "background": bg_color,
@@ -179,7 +251,10 @@ def update_marquee_settings_ui():
     })
 
 def display_marquee(text, settings, key_suffix=""):
-    """Show marquee text with style from settings."""
+    """
+    🎉 Show a marquee text with style from the marquee settings.
+    Automatically truncates text to ~280 chars to avoid overflow.
+    """
     truncated_text = text[:280] + "..." if len(text) > 280 else text
     streamlit_marquee(
         content=truncated_text,
@@ -189,7 +264,10 @@ def display_marquee(text, settings, key_suffix=""):
     st.write("")
 
 def get_high_info_terms(text: str, top_n=10) -> list:
-    """Extract top_n freq words or bigrams (excluding stopwords)."""
+    """
+    📌 Extract top_n frequent words & bigrams (excluding common stopwords).
+    Useful for generating short descriptive keywords from Q/A content.
+    """
     stop_words = set(['the', 'a', 'an', 'and', 'or', 'but', 'in', 'on', 'at', 'to', 'for', 'of', 'with'])
     words = re.findall(r'\b\w+(?:-\w+)*\b', text.lower())
     bi_grams = [' '.join(pair) for pair in zip(words, words[1:])]
@@ -199,22 +277,24 @@ def get_high_info_terms(text: str, top_n=10) -> list:
     return [term for term, freq in counter.most_common(top_n)]
 
 def clean_text_for_filename(text: str) -> str:
-    """Remove special chars, short words, etc. for filenames."""
+    """
+    🏷️ Remove special chars & short unhelpful words from text for safer filenames.
+    Returns a lowercased, underscore-joined token string.
+    """
     text = text.lower()
     text = re.sub(r'[^\w\s-]', '', text)
     words = text.split()
-    # remove short or unhelpful words
     stop_short = set(['the', 'and', 'for', 'with', 'this', 'that', 'ai', 'library'])
     filtered = [w for w in words if len(w) > 3 and w not in stop_short]
     return '_'.join(filtered)[:200]
 
 def generate_filename(prompt, response, file_type="md", max_length=200):
     """
-    Generate a shortened filename by:
-      1) extracting high-info terms,
-      2) snippet from prompt+response,
-      3) remove duplicates,
-      4) truncate if needed.
+    📁 Create a shortened filename based on prompt+response content:
+      1) Extract top info terms,
+      2) Combine snippet from prompt+response,
+      3) Remove duplicates,
+      4) Truncate if needed.
     """
     prefix = format_timestamp_prefix() + "_"
     combined_text = (prompt + " " + response)[:200]  
@@ -222,7 +302,7 @@ def generate_filename(prompt, response, file_type="md", max_length=200):
     snippet = (prompt[:40] + " " + response[:40]).strip()
     snippet_cleaned = clean_text_for_filename(snippet)
     
-    # remove duplicates
+    # Remove duplicates
     name_parts = info_terms + [snippet_cleaned]
     seen = set()
     unique_parts = []
@@ -239,88 +319,160 @@ def generate_filename(prompt, response, file_type="md", max_length=200):
     return f"{prefix}{full_name}.{file_type}"
 
 def create_file(prompt, response, file_type="md"):
-    """Create a text file from prompt + response with sanitized filename."""
+    """
+    📝 Create a text file from prompt + response with a sanitized filename.
+    Returns the created filename.
+    """
     filename = generate_filename(prompt.strip(), response.strip(), file_type)
     with open(filename, 'w', encoding='utf-8') as f:
         f.write(prompt + "\n\n" + response)
     return filename
 
-def get_download_link(file, file_type="zip"):
-    """
-    Convert a file to base64 and return an HTML link for download.
-    """
-    with open(file, "rb") as f:
-        b64 = base64.b64encode(f.read()).decode()
-    if file_type == "zip":
-        return f'<a href="data:application/zip;base64,{b64}" download="{os.path.basename(file)}">📂 Download {os.path.basename(file)}</a>'
-    elif file_type == "mp3":
-        return f'<a href="data:audio/mpeg;base64,{b64}" download="{os.path.basename(file)}">🎵 Download {os.path.basename(file)}</a>'
-    elif file_type == "wav":
-        return f'<a href="data:audio/wav;base64,{b64}" download="{os.path.basename(file)}">🔊 Download {os.path.basename(file)}</a>'
-    elif file_type == "md":
-        return f'<a href="data:text/markdown;base64,{b64}" download="{os.path.basename(file)}">📝 Download {os.path.basename(file)}</a>'
-    else:
-        return f'<a href="data:application/octet-stream;base64,{b64}" download="{os.path.basename(file)}">Download {os.path.basename(file)}</a>'
+# ─────────────────────────────────────────────────────────
+# 4. OPTIMIZED AUDIO GENERATION (ASYNC TTS + CACHING)
+# ─────────────────────────────────────────────────────────
 
 def clean_for_speech(text: str) -> str:
-    """Clean up text for TTS output."""
-    text = text.replace("\n", " ")
-    text = text.replace("</s>", " ")
-    text = text.replace("#", "")
-    text = re.sub(r"\(https?:\/\/[^\)]+\)", "", text)
-    text = re.sub(r"\s+", " ", text).strip()
-    return text
-
-async def edge_tts_generate_audio(text, voice="en-US-AriaNeural", rate=0, pitch=0, file_format="mp3"):
-    """Async TTS generation with edge-tts library."""
-    text = clean_for_speech(text)
-    if not text.strip():
-        return None
-    rate_str = f"{rate:+d}%"
-    pitch_str = f"{pitch:+d}Hz"
-    communicate = edge_tts.Communicate(text, voice, rate=rate_str, pitch=pitch_str)
-    out_fn = generate_filename(text, text, file_type=file_format)
-    await communicate.save(out_fn)
-    return out_fn
-
-def speak_with_edge_tts(text, voice="en-US-AriaNeural", rate=0, pitch=0, file_format="mp3"):
-    """Wrapper for the async TTS generate call."""
-    return asyncio.run(edge_tts_generate_audio(text, voice, rate, pitch, file_format))
-
-def play_and_download_audio(file_path, file_type="mp3"):
-    """Streamlit audio + a quick download link."""
-    if file_path and os.path.exists(file_path):
-        st.audio(file_path)
-        dl_link = get_download_link(file_path, file_type=file_type)
-        st.markdown(dl_link, unsafe_allow_html=True)
-
-def save_qa_with_audio(question, answer, voice=None):
-    """Save Q&A to markdown and also generate audio."""
-    if not voice:
-        voice = st.session_state['tts_voice']
+    """
+    🔉 Clean up text for TTS output with enhanced cleaning.
+    Removes markdown, code blocks, links, etc.
+    """
+    with PerformanceTimer("text_cleaning"):
+        # Remove markdown headers
+        text = re.sub(r'#+ ', '', text)
+        # Remove link formats [text](url)
+        text = re.sub(r'\[([^\]]+)\]\([^\)]+\)', r'\1', text)
+        # Remove emphasis markers (*, _, ~, `)
+        text = re.sub(r'[*_~`]', '', text)
+        # Remove code blocks
+        text = re.sub(r'```[\s\S]*?```', '', text)
+        text = re.sub(r'`[^`]*`', '', text)
+        # Remove excess whitespace
+        text = re.sub(r'\s+', ' ', text).replace("\n", " ")
+        # Remove hidden S tokens
+        text = text.replace("</s>", " ")
+        # Remove URLs
+        text = re.sub(r'https?://\S+', '', text)
+        text = re.sub(r'\(https?://[^\)]+\)', '', text)
+        text = text.strip()
+        return text
+
+async def async_edge_tts_generate(
+    text: str,
+    voice: str,
+    rate: int = 0,
+    pitch: int = 0,
+    file_format: str = "mp3"
+) -> Tuple[Optional[str], float]:
+    """
+    🎶 Asynchronous TTS generation with caching and performance tracking.
+    Returns (filename, generation_time).
+    """
+    with PerformanceTimer("tts_generation") as timer:
+        # ▶ Clean & validate text
+        text = clean_for_speech(text)
+        if not text.strip():
+            return None, 0
+        
+        # ▶ Check cache (avoid regenerating the same TTS)
+        cache_key = f"{text[:100]}_{voice}_{rate}_{pitch}_{file_format}"
+        if cache_key in st.session_state['audio_cache']:
+            return st.session_state['audio_cache'][cache_key], 0
+        
+        try:
+            # ▶ Generate audio
+            rate_str = f"{rate:+d}%"
+            pitch_str = f"{pitch:+d}Hz"
+            communicate = edge_tts.Communicate(text, voice, rate=rate_str, pitch=pitch_str)
+            
+            # ▶ Generate unique filename
+            timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+            filename = f"audio_{timestamp}_{random.randint(1000, 9999)}.{file_format}"
+            
+            # ▶ Save audio file
+            await communicate.save(filename)
+            
+            # ▶ Store in cache
+            st.session_state['audio_cache'][cache_key] = filename
+            
+            # ▶ Return path + timing
+            return filename, time.time() - timer.start_time
+        
+        except Exception as e:
+            st.error(f"❌ Error generating audio: {str(e)}")
+            return None, 0
+
+async def async_save_qa_with_audio(
+    question: str,
+    answer: str,
+    voice: Optional[str] = None
+) -> Tuple[str, Optional[str], float, float]:
+    """
+    📝 Asynchronously save Q&A to markdown, then generate audio if enabled.
+    Returns (md_file, audio_file, md_time, audio_time).
+    """
+    voice = voice or st.session_state['tts_voice']
     
-    combined_text = f"# Question\n{question}\n\n# Answer\n{answer}"
-    md_file = create_file(question, answer, "md")
-    audio_text = f"{question}\n\nAnswer: {answer}"
-    audio_file = speak_with_edge_tts(
-        audio_text,
-        voice=voice,
-        file_format=st.session_state['audio_format']
-    )
-    return md_file, audio_file
+    with PerformanceTimer("qa_save") as timer:
+        # ▶ Save Q/A as markdown
+        md_start = time.time()
+        md_file = create_file(question, answer, "md")
+        md_time = time.time() - md_start
+        
+        # ▶ Generate audio (if globally enabled)
+        audio_file = None
+        audio_time = 0
+        if st.session_state['enable_audio']:
+            audio_text = f"{question}\n\nAnswer: {answer}"
+            audio_file, audio_time = await async_edge_tts_generate(
+                audio_text,
+                voice=voice,
+                file_format=st.session_state['audio_format']
+            )
+        
+        return md_file, audio_file, md_time, audio_time
+
+def create_download_link_with_cache(file_path: str, file_type: str = "mp3") -> str:
+    """
+    ⬇️ Create a download link for a file with caching & error handling.
+    """
+    with PerformanceTimer("download_link_generation"):
+        cache_key = f"dl_{file_path}"
+        if cache_key in st.session_state['download_link_cache']:
+            return st.session_state['download_link_cache'][cache_key]
+        
+        try:
+            with open(file_path, "rb") as f:
+                b64 = base64.b64encode(f.read()).decode()
+            filename = os.path.basename(file_path)
+            
+            if file_type == "mp3":
+                link = f'<a href="data:audio/mpeg;base64,{b64}" download="{filename}">🎵 Download {filename}</a>'
+            elif file_type == "wav":
+                link = f'<a href="data:audio/wav;base64,{b64}" download="{filename}">🔊 Download {filename}</a>'
+            elif file_type == "md":
+                link = f'<a href="data:text/markdown;base64,{b64}" download="{filename}">📝 Download {filename}</a>'
+            else:
+                link = f'<a href="data:application/octet-stream;base64,{b64}" download="{filename}">⬇️ Download {filename}</a>'
+            
+            st.session_state['download_link_cache'][cache_key] = link
+            return link
+        
+        except Exception as e:
+            st.error(f"❌ Error creating download link: {str(e)}")
+            return ""
 
 # ─────────────────────────────────────────────────────────
-# 3. PAPER PARSING & DISPLAY
+# 5. RESEARCH / ARXIV FUNCTIONS
 # ─────────────────────────────────────────────────────────
 
 def parse_arxiv_refs(ref_text: str):
     """
-    Given a multi-line markdown with arxiv references, parse them into
-    a list of dicts: {date, title, url, authors, summary, ...}.
+    📜 Given a multi-line markdown with Arxiv references,
+    parse them into a list of dicts: {date, title, url, authors, summary}.
     """
     if not ref_text:
         return []
-
     results = []
     current_paper = {}
     lines = ref_text.split('\n')
@@ -349,10 +501,9 @@ def parse_arxiv_refs(ref_text: str):
                     'download_base64': '',
                 }
             except Exception as e:
-                st.warning(f"Error parsing paper header: {str(e)}")
+                st.warning(f"⚠️ Error parsing paper header: {str(e)}")
                 current_paper = {}
                 continue
-        
         elif current_paper:
             # If authors not set, fill it; otherwise, fill summary
             if not current_paper['authors']:
@@ -369,47 +520,47 @@ def parse_arxiv_refs(ref_text: str):
     return results[:20]
 
 def create_paper_links_md(papers):
-    """Creates a minimal .md content linking to each paper's arxiv URL."""
+    """
+    🔗 Create a minimal .md content linking to each paper's Arxiv URL.
+    """
     lines = ["# Paper Links\n"]
     for i, p in enumerate(papers, start=1):
         lines.append(f"{i}. **{p['title']}** — [Arxiv]({p['url']})")
     return "\n".join(lines)
 
-def create_paper_audio_files(papers, input_question):
+async def create_paper_audio_files(papers, input_question):
     """
-    For each paper, generate TTS audio summary, store the path in `paper['full_audio']`,
-    and also store a base64 link for stable downloading.
+    🎧 For each paper, generate TTS audio summary and store the path in `paper['full_audio']`.
+    Also creates a base64 download link in `paper['download_base64']`.
     """
     for paper in papers:
         try:
             audio_text = f"{paper['title']} by {paper['authors']}. {paper['summary']}"
             audio_text = clean_for_speech(audio_text)
             file_format = st.session_state['audio_format']
-            audio_file = speak_with_edge_tts(
+            audio_file, _ = await async_edge_tts_generate(
                 audio_text, 
                 voice=st.session_state['tts_voice'], 
                 file_format=file_format
             )
             paper['full_audio'] = audio_file
-
+            
             if audio_file:
-                with open(audio_file, "rb") as af:
-                    b64_data = base64.b64encode(af.read()).decode()
-                download_filename = os.path.basename(audio_file)
-                mime_type = "mpeg" if file_format == "mp3" else "wav"
-                paper['download_base64'] = (
-                    f'<a href="data:audio/{mime_type};base64,{b64_data}" '
-                    f'download="{download_filename}">🎵 Download {download_filename}</a>'
-                )
+                # Convert to base64 link
+                ext = file_format
+                download_link = create_download_link_with_cache(audio_file, file_type=ext)
+                paper['download_base64'] = download_link
 
         except Exception as e:
-            st.warning(f"Error processing paper {paper['title']}: {str(e)}")
+            st.warning(f"⚠️ Error processing paper {paper['title']}: {str(e)}")
             paper['full_audio'] = None
             paper['download_base64'] = ''
 
 def display_papers(papers, marquee_settings):
-    """Display paper info in the main area with marquee + expanders + audio."""
-    st.write("## Research Papers")
+    """
+    📑 Display paper info in the main area with marquee + expanders + audio.
+    """
+    st.write("## 🔎 Research Papers")
     for i, paper in enumerate(papers, start=1):
         marquee_text = f"📄 {paper['title']} | 👤 {paper['authors'][:120]} | 📝 {paper['summary'][:200]}"
         display_marquee(marquee_text, marquee_settings, key_suffix=f"paper_{i}")
@@ -419,13 +570,15 @@ def display_papers(papers, marquee_settings):
             st.markdown(f"*Authors:* {paper['authors']}")
             st.markdown(paper['summary'])
             if paper.get('full_audio'):
-                st.write("📚 Paper Audio")
+                st.write("📚 **Paper Audio**")
                 st.audio(paper['full_audio'])
                 if paper['download_base64']:
                     st.markdown(paper['download_base64'], unsafe_allow_html=True)
 
 def display_papers_in_sidebar(papers):
-    """Mirrors the paper listing in the sidebar with expanders, audio, etc."""
+    """
+    🔎 Mirrors the paper listing in the sidebar with expanders, audio, etc.
+    """
     st.sidebar.title("🎶 Papers & Audio")
     for i, paper in enumerate(papers, start=1):
         with st.sidebar.expander(f"{i}. {paper['title']}"):
@@ -439,13 +592,13 @@ def display_papers_in_sidebar(papers):
                 st.markdown(f"**Summary:** {paper['summary'][:300]}...")
 
 # ─────────────────────────────────────────────────────────
-# 4. ZIP FUNCTION
+# 6. ZIP FUNCTION
 # ─────────────────────────────────────────────────────────
 
 def create_zip_of_files(md_files, mp3_files, wav_files, input_question):
     """
-    Zip up all relevant files, limiting the final zip name to ~20 chars
-    to avoid overly long base64 strings. 
+    📦 Zip up all relevant files, generating a short name from high-info terms.
+    Returns the zip filename if created, else None.
     """
     md_files = [f for f in md_files if os.path.basename(f).lower() != 'readme.md']
     all_files = md_files + mp3_files + wav_files
@@ -455,7 +608,7 @@ def create_zip_of_files(md_files, mp3_files, wav_files, input_question):
     all_content = []
     for f in all_files:
         if f.endswith('.md'):
-            with open(f, 'r', encoding='utf-8') as file:
+            with open(f, "r", encoding='utf-8') as file:
                 all_content.append(file.read())
         elif f.endswith('.mp3') or f.endswith('.wav'):
             basename = os.path.splitext(os.path.basename(f))[0]
@@ -476,81 +629,60 @@ def create_zip_of_files(md_files, mp3_files, wav_files, input_question):
     return short_zip_name
 
 # ─────────────────────────────────────────────────────────
-# 5. MAIN LOGIC: AI LOOKUP & VOICE INPUT
+# 7. MAIN AI LOGIC: LOOKUP & TAB HANDLERS
 # ─────────────────────────────────────────────────────────
 
-def perform_ai_lookup(q, vocal_summary=True, extended_refs=False, 
-                     titles_summary=True, full_audio=False):
-    """Main routine that uses Anthropic (Claude) + Gradio ArXiv RAG pipeline."""
-    start = time.time()
-    ai_constitution = """
-    You are a talented AI coder and songwriter...
-    """
-
-    # --- 1) Claude API
-    client = anthropic.Anthropic(api_key=anthropic_key)
-    user_input = q
-    response = client.messages.create(
-        model="claude-3-sonnet-20240229",
-        max_tokens=1000,
-        messages=[
-            {"role": "user", "content": user_input}
-        ])
-    st.write("Claude's reply 🧠:")
-    st.markdown(response.content[0].text)
-
-    # Save & produce audio
-    result = response.content[0].text
-    create_file(q, result) 
-    md_file, audio_file = save_qa_with_audio(q, result)
-    st.subheader("📝 Main Response Audio")
-    play_and_download_audio(audio_file, st.session_state['audio_format'])
-    
-    # --- 2) Arxiv RAG
-    searchRAG=False
-    if searchRAG:
-        st.write("Arxiv's AI this Evening is Mixtral 8x7B...")
-        client = Client("awacke1/Arxiv-Paper-Search-And-QA-RAG-Pattern")
-        refs = client.predict(
-            q, 
-            10, 
-            "Semantic Search", 
-            "mistralai/Mixtral-8x7B-Instruct-v0.1",
-            api_name="/update_with_rag_md"
-        )[0]
-        r2 = client.predict(
-            q, 
-            "mistralai/Mixtral-8x7B-Instruct-v0.1", 
-            True, 
-            api_name="/ask_llm"
+def perform_ai_lookup(
+    q,
+    vocal_summary=True,
+    extended_refs=False,
+    titles_summary=True,
+    full_audio=False
+):
+    """
+    🔮 Main routine that uses Anthropic (Claude) + optional Gradio ArXiv RAG pipeline.
+    Currently demonstrates calling Anthropic and returning the text.
+    """
+    with PerformanceTimer("ai_lookup"):
+        start = time.time()
+        
+        # ▶ Example call to Anthropic (Claude)
+        client = anthropic.Anthropic(api_key=anthropic_key)
+        user_input = q
+        
+        # Here we do a minimal prompt, just to show the call
+        # (You can enhance your prompt engineering as needed)
+        response = client.completions.create(
+            model="claude-2",
+            max_tokens_to_sample=512,
+            prompt=f"{anthropic.HUMAN_PROMPT} {user_input}{anthropic.AI_PROMPT}"
         )
-        result = f"### 🔎 {q}\n\n{r2}\n\n{refs}"
-        md_file, audio_file = save_qa_with_audio(q, result)
-    
-        # --- 3) Parse + handle papers
-        papers = parse_arxiv_refs(refs)
-        if papers:
-            # Create minimal links page first
-            paper_links = create_paper_links_md(papers)
-            links_file = create_file(q, paper_links, "md")
-            st.markdown(paper_links)
-    
-            # Then create audio for each paper
-            create_paper_audio_files(papers, input_question=q)
-            display_papers(papers, get_marquee_settings())
-            display_papers_in_sidebar(papers)
-        else:
-            st.warning("No papers found in the response.")
-
-    elapsed = time.time() - start
-    st.write(f"**Total Elapsed:** {elapsed:.2f} s")
-    return result
+        
+        result_text = response.completion.strip()
+        
+        # ▶ Print and store
+        st.write("### Claude's reply 🧠:")
+        st.markdown(result_text)
+        
+        # ▶ We'll add to the chat history
+        st.session_state.chat_history.append({"user": q, "claude": result_text})
+        
+        # ▶ Return final text
+        end = time.time()
+        st.write(f"**Elapsed:** {end - start:.2f}s")
+        
+    return result_text
 
-def process_voice_input(text):
-    """When user sends voice query, we run the AI lookup + Q&A with audio."""
+async def process_voice_input(text):
+    """
+    🎤 When user sends a voice query, we run the AI lookup + Q/A with audio.
+    Then we store the resulting markdown & audio in session or disk.
+    """
     if not text:
         return
     st.subheader("🔍 Search Results")
+    
+    # ▶ Call AI
     result = perform_ai_lookup(
         text, 
         vocal_summary=True,
@@ -558,25 +690,98 @@ def process_voice_input(text):
         titles_summary=True,
         full_audio=True
     )
-    md_file, audio_file = save_qa_with_audio(text, result)
+    
+    # ▶ Save Q&A as Markdown + audio (async)
+    md_file, audio_file, md_time, audio_time = await async_save_qa_with_audio(text, result)
+
     st.subheader("📝 Generated Files")
-    st.write(f"Markdown: {md_file}")
-    st.write(f"Audio: {audio_file}")
-    play_and_download_audio(audio_file, st.session_state['audio_format'])
+    st.write(f"**Markdown:** {md_file} (saved in {md_time:.2f}s)")
+    if audio_file:
+        st.write(f"**Audio:** {audio_file} (generated in {audio_time:.2f}s)")
+        st.audio(audio_file)
+        dl_link = create_download_link_with_cache(audio_file, file_type=st.session_state['audio_format'])
+        st.markdown(dl_link, unsafe_allow_html=True)
+
+def display_voice_tab():
+    """
+    🎙️ Display the voice input tab with TTS settings and real-time usage.
+    """
+    st.subheader("🎤 Voice Input")
+    
+    # ▶ Voice Settings
+    st.markdown("### 🎤 Voice Settings")
+    caption_female = 'Top: 🌸 **Aria** – 🎶 **Jenny** – 🌺 **Sonia** – 🌌 **Natasha** – 🌷 **Clara**'
+    caption_male   = 'Bottom: 🌟 **Guy** – 🛠️ **Ryan** – 🎻 **William** – 🌟 **Liam**'
+    
+    # Optionally, replace with your own local image or comment out
+    # st.sidebar.image('Group Picture - Voices.png', caption=caption_female + ' | ' + caption_male)
+    
+    st.sidebar.markdown("""
+    # 🎙️ Voice Character Agent Selector 🎭
+    *Female Voices*:
+    - 🌸 **Aria** – Elegant, creative storytelling  
+    - 🎶 **Jenny** – Friendly, conversational  
+    - 🌺 **Sonia** – Bold, confident  
+    - 🌌 **Natasha** – Sophisticated, mysterious  
+    - 🌷 **Clara** – Cheerful, empathetic  
+
+    *Male Voices*:
+    - 🌟 **Guy** – Authoritative, versatile  
+    - 🛠️ **Ryan** – Approachable, casual  
+    - 🎻 **William** – Classic, scholarly  
+    - 🌟 **Liam** – Energetic, engaging
+    """)
+    
+    selected_voice = st.selectbox(
+        "👄 Select TTS Voice:",
+        options=EDGE_TTS_VOICES,
+        index=EDGE_TTS_VOICES.index(st.session_state['tts_voice'])
+    )
+    
+    # ▶ Audio Format
+    st.markdown("### 🔊 Audio Format")
+    selected_format = st.radio(
+        "Choose Audio Format:",
+        options=["MP3", "WAV"],
+        index=0
+    )
+
+    # ▶ Update session state if changed
+    if selected_voice != st.session_state['tts_voice']:
+        st.session_state['tts_voice'] = selected_voice
+        st.experimental_rerun()
+    if selected_format.lower() != st.session_state['audio_format']:
+        st.session_state['audio_format'] = selected_format.lower()
+        st.experimental_rerun()
+
+    # ▶ Text Input
+    user_text = st.text_area("💬 Message:", height=100)
+    user_text = user_text.strip().replace('\n', ' ')
+
+    # ▶ Send Button
+    if st.button("📨 Send"):
+        # Run our process_voice_input as an async function
+        asyncio.run(process_voice_input(user_text))
+
+    # ▶ Chat History
+    st.subheader("📜 Chat History")
+    for c in st.session_state.chat_history:
+        st.write("**You:**", c["user"])
+        st.write("**Response:**", c["claude"])
 
 # ─────────────────────────────────────────────��───────────
-# 6. FILE HISTORY SIDEBAR
+# FILE HISTORY SIDEBAR
 # ─────────────────────────────────────────────────────────
 
 def display_file_history_in_sidebar():
     """
-    Shows a history of each local .md, .mp3, .wav file in descending
-    order of modification time, with quick icons and optional download links.
+    📂 Shows a history of local .md, .mp3, .wav files (newest first),
+    with quick icons and optional download links.
     """
     st.sidebar.markdown("---")
     st.sidebar.markdown("### 📂 File History")
 
-    # Gather all files
+    # ▶ Gather all files
     md_files = glob.glob("*.md")
     mp3_files = glob.glob("*.mp3")
     wav_files = glob.glob("*.wav")
@@ -586,7 +791,7 @@ def display_file_history_in_sidebar():
         st.sidebar.write("No files found.")
         return
 
-    # Sort newest first
+    # ▶ Sort newest first
     all_files = sorted(all_files, key=os.path.getmtime, reverse=True)
 
     for f in all_files:
@@ -603,39 +808,44 @@ def display_file_history_in_sidebar():
                 if len(snippet) == 200:
                     snippet += "..."
                 st.write(snippet)
-                st.markdown(get_download_link(f, file_type="md"), unsafe_allow_html=True)
+                dl_link = create_download_link_with_cache(f, file_type="md")
+                st.markdown(dl_link, unsafe_allow_html=True)
             elif ext in ["mp3","wav"]:
                 st.audio(f)
-                st.markdown(get_download_link(f, file_type=ext), unsafe_allow_html=True)
+                dl_link = create_download_link_with_cache(f, file_type=ext)
+                st.markdown(dl_link, unsafe_allow_html=True)
             else:
-                st.markdown(get_download_link(f), unsafe_allow_html=True)
+                dl_link = create_download_link_with_cache(f)
+                st.markdown(dl_link, unsafe_allow_html=True)
 
 # ─────────────────────────────────────────────────────────
-# 7. MAIN APP
+# MAIN APP
 # ─────────────────────────────────────────────────────────
 
 def main():
-    # 1) Setup marquee UI in the sidebar
+    # ▶ 1) Setup marquee UI in the sidebar
     update_marquee_settings_ui()
     marquee_settings = get_marquee_settings()
 
-    # 2) Display the marquee welcome
-    display_marquee(st.session_state['marquee_content'], 
-                    {**marquee_settings, "font-size": "28px", "lineHeight": "50px"},
-                    key_suffix="welcome")
+    # ▶ 2) Display the marquee welcome
+    display_marquee(
+        st.session_state['marquee_content'], 
+        {**marquee_settings, "font-size": "28px", "lineHeight": "50px"},
+        key_suffix="welcome"
+    )
 
-    # 3) Main action tabs
+    # ▶ 3) Main action tabs
     tab_main = st.radio("Action:", ["🎤 Voice", "📸 Media", "🔍 ArXiv", "📝 Editor"], 
                         horizontal=True)
 
-    # Example custom component usage
+    # ▶ 4) Show or hide custom component (optional example)
     mycomponent = components.declare_component("mycomponent", path="mycomponent")
-    val = mycomponent(my_input_value="Hello")
+    val = mycomponent(my_input_value="Hello from MyComponent")
 
     if val:
         val_stripped = val.replace('\\n', ' ')
         edited_input = st.text_area("✏️ Edit Input:", value=val_stripped, height=100)
-        run_option = st.selectbox("Model:", ["Arxiv"])
+        run_option = st.selectbox("Model:", ["Arxiv", "Other (demo)"])
         col1, col2 = st.columns(2)
         with col1:
             autorun = st.checkbox("⚙ AutoRun", value=True)
@@ -661,7 +871,7 @@ def main():
                                   extended_refs=False, 
                                   titles_summary=True, 
                                   full_audio=full_audio)
-    
+
     # ─────────────────────────────────────────────────────────
     # TAB: ArXiv
     # ─────────────────────────────────────────────────────────
@@ -678,8 +888,11 @@ def main():
         
         if q and st.button("🔍Run"):
             st.session_state.last_query = q
-            result = perform_ai_lookup(q, vocal_summary=vocal_summary, extended_refs=extended_refs, 
-                                       titles_summary=titles_summary, full_audio=full_audio)
+            result = perform_ai_lookup(q, 
+                                       vocal_summary=vocal_summary, 
+                                       extended_refs=extended_refs, 
+                                       titles_summary=titles_summary, 
+                                       full_audio=full_audio)
             if full_transcript:
                 create_file(q, result, "md")
 
@@ -687,52 +900,16 @@ def main():
     # TAB: Voice
     # ─────────────────────────────────────────────────────────
     elif tab_main == "🎤 Voice":
-        st.subheader("🎤 Voice Input")
-        
-        st.markdown("### 🎤 Voice Settings")
-        selected_voice = st.selectbox(
-            "Select TTS Voice:",
-            options=EDGE_TTS_VOICES,
-            index=EDGE_TTS_VOICES.index(st.session_state['tts_voice'])
-        )
-        
-        st.markdown("### 🔊 Audio Format")
-        selected_format = st.radio(
-            "Choose Audio Format:",
-            options=["MP3", "WAV"],
-            index=0
-        )
-
-        # Update session state if voice/format changes
-        if selected_voice != st.session_state['tts_voice']:
-            st.session_state['tts_voice'] = selected_voice
-            st.rerun()
-        if selected_format.lower() != st.session_state['audio_format']:
-            st.session_state['audio_format'] = selected_format.lower()
-            st.rerun()
-
-        # Input text
-        user_text = st.text_area("💬 Message:", height=100)
-        user_text = user_text.strip().replace('\n', ' ')
-
-        if st.button("📨 Send"):
-            process_voice_input(user_text)
-
-        st.subheader("📜 Chat History")
-        for c in st.session_state.chat_history:
-            st.write("**You:**", c["user"])
-            st.write("**Response:**", c["claude"])
+        display_voice_tab()
 
     # ─────────────────────────────────────────────────────────
     # TAB: Media
     # ─────────────────────────────────────────────────────────
     elif tab_main == "📸 Media":
         st.header("📸 Media Gallery")
-
-        # By default, show audio first
         tabs = st.tabs(["🎵 Audio", "🖼 Images", "🎥 Video"])
         
-        # AUDIO sub-tab
+        # ▶ AUDIO sub-tab
         with tabs[0]:
             st.subheader("🎵 Audio Files")
             audio_files = glob.glob("*.mp3") + glob.glob("*.wav")
@@ -741,12 +918,12 @@ def main():
                     with st.expander(os.path.basename(a)):
                         st.audio(a)
                         ext = os.path.splitext(a)[1].replace('.', '')
-                        dl_link = get_download_link(a, file_type=ext)
+                        dl_link = create_download_link_with_cache(a, file_type=ext)
                         st.markdown(dl_link, unsafe_allow_html=True)
             else:
                 st.write("No audio files found.")
         
-        # IMAGES sub-tab
+        # ▶ IMAGES sub-tab
         with tabs[1]:
             st.subheader("🖼 Image Files")
             imgs = glob.glob("*.png") + glob.glob("*.jpg") + glob.glob("*.jpeg")
@@ -759,7 +936,7 @@ def main():
             else:
                 st.write("No images found.")
         
-        # VIDEO sub-tab
+        # ▶ VIDEO sub-tab
         with tabs[2]:
             st.subheader("🎥 Video Files")
             vids = glob.glob("*.mp4") + glob.glob("*.mov") + glob.glob("*.avi")
@@ -774,14 +951,16 @@ def main():
     # TAB: Editor
     # ─────────────────────────────────────────────────────────
     elif tab_main == "📝 Editor":
-        st.write("Select or create a file to edit. (Currently minimal demo)")
+        st.write("### 📝 File Editor (Minimal Demo)")
+        st.write("Select or create a file to edit. More advanced features can be added as needed.")
 
     # ─────────────────────────────────────────────────────────
-    # SIDEBAR: FILE HISTORY
+    # SIDEBAR: FILE HISTORY + PERFORMANCE METRICS
     # ─────────────────────────────────────────────────────────
     display_file_history_in_sidebar()
+    log_performance_metrics()
 
-    # Some light CSS styling
+    # ▶ Some light CSS styling
     st.markdown("""
     <style>
         .main { background: linear-gradient(to right, #1a1a1a, #2d2d2d); color: #fff; }
@@ -790,10 +969,14 @@ def main():
     </style>
     """, unsafe_allow_html=True)
 
-    # Rerun if needed
+    # ▶ Rerun if needed
     if st.session_state.should_rerun:
         st.session_state.should_rerun = False
-        st.rerun()
+        st.experimental_rerun()
+
+# ────────────────────────────────────────────���────────────
+# 8. RUN APP
+# ─────────────────────────────────────────────────────────
 
 if __name__ == "__main__":
     main()