🤖 Advanced Sentiment Analysis Platform ✨

An intelligent, multi-modal sentiment analysis tool built with Streamlit, Transformers, and Whisper. This platform doesn't just read text; it understands emotion, sarcasm, and even the tone of voice from video and audio files.

---

🚀 Key Features

This platform provides a deep, comprehensive understanding of sentiment by combining multiple AI models.

• ✨ Multi-Modal Analysis: Analyze text, audio (.mp3, .wav), and video (.mp4) files.
• 📈 Hybrid Scoring: Combines the power of RoBERTa (a Transformer model) with the speed of VADER (a rule-based model) for a more accurate and nuanced sentiment score.
• 🎯 Aspect-Based Sentiment: Goes beyond a simple overall score. It finds what you're talking about (e.g., "camera," "battery life") and analyzes the sentiment for each specific aspect.
• 🎬 File & YouTube Analysis: Upload your own files or just paste a YouTube URL to download, transcribe, and analyze the content.
• 🗣️ Speech Emotion Recognition: It's not just what you say, but how you say it. The tool analyzes the audio's tone to detect emotions like joy, sadness, or anger.
• 🧐 Sarcasm Detection: A dedicated model looks for irony and sarcasm, warning you if the positive words might not mean what you think.
• 🤖 Sentiment-Aware Chatbot: Have a conversation with an AI that understands the emotion behind your messages.

---

🛠️ How It Works (The Tech)

This app's power comes from a "committee" of AI models that work together.

1. Input: You provide text, a file, or a URL.
2. Transcription: If it's audio or video, OpenAI's Whisper model creates a highly accurate text transcript.
3. Sentiment Analysis:
   • RoBERTa and VADER analyze the text for a hybrid sentiment score.
   • BERT (a 43-label emotion model) detects emotions like joy, anger, or admiration.
   • A Sarcasm-detection model flags potential irony.
4. Speech Analysis: A Wav2Vec2 model analyzes the raw audio (from the file or video) to find the emotion in the tone of voice.
5. NLP: SpaCy is used to break down sentences and find the "aspects" (noun chunks) for analysis.
6. Output: All this information is presented in a clean, interactive dashboard with charts and detailed breakdowns.

---

🏛️ Project Structure

This project has been refactored for clarity, maintainability, and scalability. The logic is now separated from the user interface.

• app.py: The frontend of the application. This file contains all the Streamlit code responsible for the user interface (UI), such as pages, buttons, charts, and layout.
• analysis_logic.py: The backend "brain" of the application. This file contains all the core data processing and AI functions (model loading, text analysis, web scraping, video transcription, etc.).
• requirements.txt: A list of all required Python packages.

---

🛠️ How to Run

1. Clone the Repository:

   git clone [https://github.com/Dhy4n-117/Sentiment-Analysis.git]
   cd Sentiment-Analysis

2. Install Prerequisites:

   • This project requires FFmpeg (for video/audio) and PortAudio (for voice recording).

   • On Windows:

     • Download and install FFmpeg from gyan.dev (get the ffmpeg-release-full.7z file).
     • Extract it and add the bin folder to your Windows PATH environment variable.
     • pyaudio (for voice) is installed via pip, but it may require you to install "Microsoft C++ Build Tools" if it fails.

   • On Linux (Debian/Ubuntu):

     • Run the following command to install both prerequisites from the terminal:

       sudo apt-get update && sudo apt-get install -y ffmpeg portaudio19-dev

   • On macOS:

     • Use Homebrew to install:

       brew install ffmpeg portaudio

3. Create and Activate a Virtual Environment:

   python -m venv venv
   .\venv\Scripts\activate

4. Install Python Dependencies:

   pip install -r requirements.txt

5. Download SpaCy Model:

   python -m spacy download en_core_web_sm

6. Run the App:

   streamlit run app.py

---