works cool

README.md

@@ -15,7 +15,8 @@ tags:
   - ai
   - radio
   - music
-  - gemini
+  - nebius
+  - gpt-oss-120b
   - elevenlabs
   - llamaindex
   - rag
@@ -72,7 +73,7 @@ ai_radio/
 - Perfect short-form content between segments
 
 ### 🤖 AI Radio Host
-- Charismatic AI host powered by Google Gemini
+- Charismatic AI host powered by Nebius GPT-OSS-120B
 - Personalized greetings and interactions
 - Smooth transitions between segments
 
@@ -102,7 +103,7 @@ This app fulfills all requirements for **Track 2: MCP in Action**:
 ## 🛠️ Technology Stack
 
 - **Gradio**: Interactive web interface
-- **Google Gemini (Gemini Pro)**: LLM for content generation, host commentary, and reasoning
+- **Nebius GPT-OSS-120B** (OpenAI-compatible): LLM for content generation, host commentary, and reasoning
 - **ElevenLabs**: High-quality text-to-speech for voice generation
 - **LlamaIndex**: RAG system for personalized recommendations and user preference management
 - **MCP (Model Context Protocol)**: Structured tool servers for modular functionality
@@ -114,7 +115,7 @@ This app fulfills all requirements for **Track 2: MCP in Action**:
 ### Prerequisites
 
 - Python 3.9+
-- Google Gemini API key (get from [Google AI Studio](https://makersuite.google.com/app/apikey))
+- Nebius API key (for GPT-OSS-120B LLM)
 - ElevenLabs API key (provided)
 
 ### Installation
@@ -130,9 +131,9 @@ cd ai_radio
 pip install -r requirements.txt
 ```
 
-3. Set up your Gemini API key in `config.py`:
+3. Set up your Nebius API key in `config.py`:
 ```python
-google_api_key: str = "your-gemini-api-key-here"
+nebius_api_key: str = "your-nebius-api-key-here"
 ```
 
 4. Run the app:
@@ -191,7 +192,7 @@ The **RadioAgent** is an autonomous AI agent that:
 
 1. **Plans**: Analyzes user preferences and creates a balanced show plan with music, news, podcasts, and stories
 2. **Reasons**: Uses the RAG system to make intelligent decisions about content selection
-3. **Executes**: Generates content using Gemini LLM and delivers it via ElevenLabs TTS
+3. **Executes**: Generates content using Nebius GPT-OSS-120B LLM and delivers it via ElevenLabs TTS
 
 ### MCP Server Architecture
 
@@ -226,7 +227,7 @@ The agent uses intelligent planning to create a balanced show:
 1. **Analyze preferences**: Load user preferences from RAG system
 2. **Calculate distribution**: Determine segment ratios (50% music, 20% news, 20% podcasts, 10% stories)
 3. **Generate segments**: Use MCP servers to fetch content for each segment
-4. **Add personality**: Generate host commentary using Gemini LLM
+4. **Add personality**: Generate host commentary using Nebius GPT-OSS-120B LLM
 5. **Execute**: Convert text to speech and play audio
 
 ## 🔊 Audio Generation
@@ -261,7 +262,7 @@ MIT License - feel free to use and modify as needed.
 ## 🙏 Acknowledgments
 
 - **MCP Team** for the amazing protocol and competition
-- **Google** for Gemini API
+- **Nebius** for GPT-OSS-120B API
 - **ElevenLabs** for text-to-speech technology
 - **LlamaIndex** for RAG capabilities
 - **Gradio** for the beautiful UI framework

requirements.txt

@@ -1,11 +1,8 @@
 gradio==4.44.0
 openai>=1.0.0
-# google-generativeai>=0.5.2,<0.6.0  # Commented out - using Nebius instead
 elevenlabs==1.10.0
 llama-index==0.11.20
 llama-index-llms-openai>=0.1.5
-# llama-index-llms-gemini==0.3.4  # Commented out - using Nebius instead
-# llama-index-embeddings-gemini==0.2.2  # Commented out - using Nebius instead
 requests==2.32.3
 python-dotenv==1.0.1
 pydantic==2.9.2

src/app.py

@@ -577,10 +577,20 @@ def play_next_segment():
     return segment_info, host_audio_file, music_player_html, progress, get_now_playing(segment), display_script
 
 def stop_radio():
-    """Stop the radio stream"""
+    """Stop the radio stream - clears audio and music player"""
     radio_state["stop_flag"] = True
     radio_state["is_playing"] = False
-    return "⏸️ Radio stopped. Press Play to resume!"
+    radio_state["planned_show"] = []
+    radio_state["current_segment_index"] = 0
+    
+    # Return status, clear audio, clear music player, reset progress, reset now playing
+    return (
+        "⏹️ Radio stopped.",
+        None,  # Clear audio
+        "",    # Clear music player HTML
+        "Stopped",
+        "📻 Ready to start"
+    )
 
 def format_segment_info(segment: Dict[str, Any]) -> str:
     """Format segment information for display"""
@@ -901,9 +911,7 @@ Your personal passphrase is:
 
 🔑 **{passphrase}**
 
-⚠️ **Save this passphrase!** You'll need it to access your account from other devices or if cookies are cleared.
-
-Your passphrase is stored in a cookie for automatic login on this browser."""
+⚠️ **Save this passphrase!** You'll need it to access your account from other devices or if cookies are cleared."""
     
     return message, user_id, passphrase
 
@@ -973,16 +981,36 @@ def get_user_stats_display() -> str:
 """
 
 def get_stats():
-    """Get listening statistics"""
-    stats = agent.rag_system.get_listening_stats()
+    """Get listening statistics for the current user"""
+    user_id = radio_state.get("user_id")
+    
+    if not user_id:
+        return """📊 **Your Listening Stats**
+
+*Log in or create an account to track your listening history!*
+"""
+    
+    # Get user-specific stats from user_memory
+    user_stats = user_memory.get_user_stats(user_id)
+    liked_tracks = user_memory.get_liked_tracks(user_id)
+    disliked_tracks = user_memory.get_disliked_tracks(user_id)
+    play_history = user_memory.get_play_history(user_id)
+    
+    # Count by type from play history
+    music_count = sum(1 for t in play_history if t.get('source') in ['youtube', 'soundcloud'])
     
     return f"""📊 **Your Listening Stats**
 
-🎧 Total Sessions: {stats['total_sessions']}
-🎵 Music Played: {stats['music_played']}
-📰 News Segments: {stats['news_heard']}
-🎙️ Podcasts: {stats['podcasts_listened']}
-📖 Stories: {stats['stories_enjoyed']}
+👤 **User ID:** {user_id}
+
+🎵 **Total Tracks Played:** {user_stats.get('total_plays', 0)}
+👍 **Liked Tracks:** {len(liked_tracks)}
+👎 **Disliked Tracks:** {len(disliked_tracks)}
+
+📜 **Recent History:** {len(play_history)} tracks
+
+*Your liked tracks have a 30% chance of playing during music segments!*
+*Disliked tracks will be skipped automatically.*
 """
 
 # Custom CSS for beautiful radio station UI
@@ -1090,9 +1118,9 @@ with gr.Blocks(css=custom_css, title="AI Radio 🎵", theme=gr.themes.Soft()) as
             gr.Markdown("---")
             gr.Markdown("""
             **💡 How it works:**
-            - Your passphrase is stored in a browser cookie for automatic login
+            - If you are here for the first time, create a new account
             - Use the same passphrase on any device to access your account
-            - If cookies are cleared, just enter your passphrase to log back in
+            - Just enter your passphrase to log back in
             """)
         
         # Tab 1: Preferences
@@ -1202,7 +1230,7 @@ with gr.Blocks(css=custom_css, title="AI Radio 🎵", theme=gr.themes.Soft()) as
             stop_btn.click(
                 fn=stop_radio,
                 inputs=[],
-                outputs=[status_text]
+                outputs=[status_text, audio_output, music_player, progress_text, now_playing]
             )
             
             # Voice input button - direct click without .then() chain
@@ -1309,8 +1337,8 @@ Your passphrase: **{passphrase}**
             outputs=[],
             js="(userId, passphrase) => { if(userId && passphrase) { localStorage.setItem('ai_radio_user_id', userId); localStorage.setItem('ai_radio_passphrase', passphrase); } }"
         ).then(
-            # Go to preferences tab
-            fn=lambda uid: gr.Tabs(selected=1) if uid else gr.Tabs(selected=0),
+            # Go to Radio Player tab after successful login
+            fn=lambda uid: gr.Tabs(selected=2) if uid else gr.Tabs(selected=0),
             inputs=[auth_user_id],
             outputs=[tabs]
         )
@@ -1403,35 +1431,64 @@ Your passphrase: **{passphrase}**
             
             ## 🌟 Features
             
-            - **🎵 Personalized Music**: Curated tracks based on your favorite genres and mood
-            - **📰 Custom News**: News updates on topics you care about
+            - **🎵 Personalized Music**: Curated tracks from YouTube based on your favorite genres and mood
+            - **📰 Custom News**: Real-time news updates on topics you care about
             - **🎙️ Podcast Recommendations**: Discover interesting podcasts matching your interests
             - **📖 AI-Generated Stories**: Entertaining stories and fun facts
-            - **🤖 AI Host**: Dynamic AI radio host that introduces segments
+            - **🤖 AI Host**: Dynamic AI radio host (Lera) that introduces segments
             - **💾 Smart Recommendations**: RAG system learns from your listening history
+            - **🔐 User Accounts**: Passphrase-based authentication with saved preferences
+            
+            ## 🧠 How It Works
+            
+            ### MCP Servers (Modular Tools)
+            
+            Three specialized **MCP (Model Context Protocol)** servers act as tools:
+            - **Music Server**: Searches YouTube for music tracks matching your preferences
+            - **News Server**: Fetches real-time news from RSS feeds
+            - **Podcast Server**: Discovers podcasts on YouTube
+            
+            ### LLM (Large Language Model)
+            
+            **Nebius GPT-OSS-120B** generates all text content:
+            - Personalized host commentary and introductions
+            - Conversational news scripts
+            - Entertaining stories and fun facts
+            - All adapted to your mood and preferences
+            
+            ### RAG System (Retrieval-Augmented Generation)
+            
+            **LlamaIndex-powered RAG** provides context-aware personalization:
+            - Stores your preferences and listening history
+            - Retrieves context for better recommendations
+            - Learns from your behavior to improve suggestions over time
             
             ## 🛠️ Technology Stack
             
             - **Gradio**: Beautiful, interactive UI
-            - **Nebius GPT-OSS-120B (OpenAI-compatible)**: LLM for content generation and host commentary
-            - **Open-source local TTS (pyttsx3)**: Offline text-to-speech for voice generation
+            - **Nebius GPT-OSS-120B** (OpenAI-compatible): LLM for content generation
+            - **ElevenLabs**: High-quality text-to-speech for voice generation
             - **LlamaIndex**: RAG system for personalized recommendations
             - **MCP Servers**: Modular tools for music, news, and podcasts
+            - **yt-dlp**: YouTube music and podcast search
+            - **YouTube/SoundCloud**: Music and podcast streaming
             
             ## 🏆 Built for MCP 1st Birthday Competition
             
             This app demonstrates:
-            - ✅ Autonomous agent behavior (planning, reasoning, execution)
-            - ✅ MCP servers as tools (music, news, podcast servers)
-            - ✅ Context engineering and RAG for personalization
-            - ✅ Gradio interface for seamless user experience
+            - ✅ **Autonomous Agent Behavior**: Planning, reasoning, and execution
+            - ✅ **MCP Servers as Tools**: Modular music, news, and podcast servers
+            - ✅ **RAG System**: Context-aware personalization with LlamaIndex
+            - ✅ **LLM Integration**: Content generation with Nebius GPT-OSS-120B
+            - ✅ **Gradio Interface**: Seamless user experience
             
             ## 📝 How to Use
             
-            1. **Set Your Preferences**: Go to the "Your Preferences" tab and tell us about yourself
-            2. **Start Radio**: Head to "Radio Player" and click "Start Radio"
-            3. **Enjoy**: Listen to your personalized show and use controls to navigate
-            4. **Track Stats**: Check "Your Stats" to see your listening history
+            1. **Create Account**: Get your unique passphrase (saved in browser)
+            2. **Set Preferences**: Choose genres, interests, and mood
+            3. **Start Radio**: Click "Generate & Play" to begin your personalized show
+            4. **Interact**: Like/dislike tracks, request songs by voice
+            5. **Track Stats**: View your listening history and statistics
             
             ---
             

src/config.py

@@ -7,7 +7,6 @@ class RadioConfig(BaseModel):
     
     # API Keys
     elevenlabs_api_key: str = "sk_2dde999f3cedf21dff7ba4671ce27f292e48ea37d30c5e4a"
-    google_api_key: str = "AIzaSyB5F9P0oDZ6fgW8GgADfwnwcg-GkHrdo74"
     llamaindex_api_key: str = "llx-WRsj0iehk2ZlSlNIenOLyyhO9X1yFT4CmJXpl0qk6hapFi01"
     nebius_api_key: str = "v1.CmQKHHN0YXRpY2tleS1lMDB0eTkxeTdwY3lxNDk5OWcSIXNlcnZpY2VhY2NvdW50LWUwMGowemtmZWpqc2E3ZHF3aDIMCKb4oskGENS9j8MBOgwIpfu6lAcQgOqAhwNAAloDZTAw.AAAAAAAAAAGEI_L5sJCQ7XR93nSzvXCPO-J3-gHjqPiRqrvkrMLeDtd-70zGWB1-c8yovnX-q7yEc1dHOnA2L8FUa3Le6X8D"
     # Nebius OpenAI-compatible endpoint and model
@@ -40,8 +39,6 @@ def get_config() -> RadioConfig:
     # Override with environment variables if available
     if api_key := os.getenv("ELEVENLABS_API_KEY"):
         config.elevenlabs_api_key = api_key
-    if api_key := os.getenv("GOOGLE_API_KEY"):
-        config.google_api_key = api_key
     if api_key := os.getenv("LLAMAINDEX_API_KEY"):
         config.llamaindex_api_key = api_key
     if api_key := os.getenv("NEBIUS_API_KEY"):

src/radio_agent.py

@@ -6,7 +6,6 @@ import os
 from typing import Dict, Any, List, Generator
 from datetime import datetime
 from openai import OpenAI
-# import google.generativeai as genai  # Commented out - using Nebius/OpenAI instead
 
 from mcp_servers.music_server import MusicMCPServer
 from mcp_servers.news_server import NewsMCPServer
@@ -50,17 +49,6 @@ class RadioAgent:
                 print("Radio agent will work in fallback mode with limited LLM features")
                 self.client = None
         
-        # # Initialize Gemini LLM with error handling (COMMENTED OUT - for future use)
-        # self.model = None
-        # if config.google_api_key:
-        #     try:
-        #         genai.configure(api_key=config.google_api_key)
-        #         self.model = genai.GenerativeModel('gemini-1.0-pro')
-        #     except Exception as e:
-        #         print(f"Warning: Could not initialize Gemini LLM: {e}")
-        #         print("Radio agent will work in fallback mode with limited LLM features")
-        #         self.model = None
-        
         # Initialize MCP Servers
         self.music_server = MusicMCPServer()
         self.news_server = NewsMCPServer()
@@ -238,14 +226,6 @@ class RadioAgent:
             except Exception as e:
                 print(f"Error generating intro: {e}")
         
-        # # Gemini fallback (COMMENTED OUT - for future use)
-        # if self.model:
-        #     try:
-        #         response = self.model.generate_content(prompt)
-        #         return response.text
-        #     except Exception as e:
-        #         print(f"Error generating intro: {e}")
-        
         # Fallback intro
         return f"Good {time_of_day}, {name}! Welcome to your personal AI Radio station. We've got an amazing show lined up for you today! Did you know that music can actually boost your productivity? That's right! So sit back, relax, and let's get this party started!"
     

src/rag_system.py

@@ -10,8 +10,6 @@ from llama_index.llms.openai import OpenAI as LlamaOpenAI
 
 # Get project root directory (parent of src/)
 PROJECT_ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
-# from llama_index.embeddings.gemini import GeminiEmbedding  # Commented out - using Nebius instead
-# from llama_index.llms.gemini import Gemini  # Commented out - using Nebius instead
 
 class RadioRAGSystem:
     """RAG system for storing and retrieving user preferences and listening history"""
@@ -43,24 +41,6 @@ class RadioRAGSystem:
             # Embeddings can be added later if needed
             self.embedding_available = False  # Disabled for now - can add OpenAI embeddings later
         
-        # # Configure LlamaIndex settings with Gemini (COMMENTED OUT - for future use)
-        # if google_api_key:
-        #     try:
-        #         Settings.llm = Gemini(api_key=google_api_key, model="models/gemini-1.0-pro")
-        #         self.llm_available = True
-        #     except Exception as e:
-        #         print(f"Warning: Could not initialize Gemini LLM: {e}")
-        #         print("RAG system will work in fallback mode without LLM features")
-        #         self.llm_available = False
-        #     
-        #     try:
-        #         Settings.embed_model = GeminiEmbedding(api_key=google_api_key, model_name="models/embedding-001")
-        #         self.embedding_available = True
-        #     except Exception as e:
-        #         print(f"Warning: Could not initialize Gemini Embeddings: {e}")
-        #         print("RAG system will work in fallback mode without embeddings")
-        #         self.embedding_available = False
-        
         # Initialize vector store
         self.vector_store = SimpleVectorStore()
         self.storage_context = StorageContext.from_defaults(vector_store=self.vector_store)

src/test_app.py

@@ -13,10 +13,10 @@ def test_imports():
         return False
     
     try:
-        import google.generativeai as genai
-        print("  ✅ Google Generative AI imported")
+        from openai import OpenAI
+        print("  ✅ OpenAI (Nebius) imported")
     except ImportError as e:
-        print(f"  ❌ Google Generative AI import failed: {e}")
+        print(f"  ❌ OpenAI import failed: {e}")
         return False
     
     try:
@@ -56,11 +56,11 @@ def test_config():
         else:
             print("  ⚠️  ElevenLabs API key missing")
         
-        if config.google_api_key:
-            print("  ✅ Google API key present")
+        if config.nebius_api_key:
+            print("  ✅ Nebius API key present")
         else:
-            print("  ⚠️  Google API key missing (REQUIRED!)")
-            print("     Please add your Gemini API key to config.py")
+            print("  ⚠️  Nebius API key missing (REQUIRED!)")
+            print("     Please add your Nebius API key to config.py")
         
         if config.llamaindex_api_key:
             print("  ✅ LlamaIndex API key present")
@@ -126,7 +126,7 @@ def test_rag_system():
         from config import get_config
         
         config = get_config()
-        rag = RadioRAGSystem(config.google_api_key)
+        rag = RadioRAGSystem(config.nebius_api_key, config.nebius_api_base, config.nebius_model)
         print("  ✅ RAG system initialized")
         
         # Test storing preferences
@@ -285,15 +285,15 @@ def main():
     if passed == total:
         print("\n🎉 All tests passed! Your AI Radio is ready to launch! 🎉")
         print("\nNext steps:")
-        print("  1. Make sure you've added your Gemini API key to config.py")
-        print("  2. Run: python app.py")
-        print("  3. Open: http://localhost:7860")
+        print("  1. Make sure you've added your Nebius API key to config.py")
+        print("  2. Run: python run.py")
+        print("  3. Open: http://localhost:7867")
         return 0
     else:
         print("\n⚠️  Some tests failed. Please fix the issues above.")
         print("\nCommon fixes:")
         print("  - Run: pip install -r requirements.txt")
-        print("  - Add your Gemini API key to config.py")
+        print("  - Add your Nebius API key to config.py")
         print("  - Check that all files are in the correct location")
         return 1
 

src/user_memory.py

@@ -270,6 +270,32 @@ class UserMemoryService:
             return []
         return self.users[user_id].get("liked_tracks", [])
     
+    def get_disliked_tracks(self, user_id: str) -> List[str]:
+        """Get user's disliked track identifiers
+        
+        Args:
+            user_id: User ID
+            
+        Returns:
+            List of disliked track identifiers (title|artist)
+        """
+        if user_id not in self.users:
+            return []
+        return self.users[user_id].get("disliked_tracks", [])
+    
+    def get_play_history(self, user_id: str) -> List[Dict[str, Any]]:
+        """Get user's play history
+        
+        Args:
+            user_id: User ID
+            
+        Returns:
+            List of played track dicts
+        """
+        if user_id not in self.users:
+            return []
+        return self.users[user_id].get("play_history", [])
+    
     def get_random_liked_track(self, user_id: str, genre: str = None) -> Optional[Dict[str, Any]]:
         """Get a random liked track, optionally filtered by genre