Lovable Prompt Creator - AI Prompt Generator

## APPLICATION OVERVIEW
The application is a web-based dashboard designed to provide real-time analysis of audio calls, specifically targeting spam and fraud detection during conversations. It aims to enhance user security by alerting them instantly about potential scams while ensuring a seamless user experience.

## CORE FEATURES
1. **Live Call Audio Capture**: Capture and stream live audio in small chunks (1-3 seconds) for real-time analysis.
2. **Real-Time Speech-to-Text**: Implement streaming speech-to-text functionality to convert audio into text during the call.
3. **Continuous Audio Analysis**: Analyze audio transcripts continuously to detect patterns indicative of scams or fraud.
4. **Dynamic Fraud Risk Scoring**: Generate and update a fraud risk score (0-100) dynamically based on detected patterns.
5. **Instant Alerts**: Trigger alerts to the user when the risk score exceeds a defined threshold, ensuring immediate awareness.
6. **Multilingual Support**: Support for multiple languages including English, Hindi, and Tamil, catering to the diverse Indian population.

## DESIGN SPECIFICATIONS
- **Visual Style**: Minimalist - The design is clean and simple, focusing on functionality with ample white space to enhance readability and usability.
- **Color Mode**: Light theme with dark text on light backgrounds to ensure high contrast and legibility.
- **Layout**: A two-column layout where the left column displays the live transcript and risk score meter, while the right column features instant alerts and additional information.
- **Typography**: Use a sans-serif font (e.g., Inter or Roboto) for clarity, with headings in a larger size for emphasis and body text in a standard size for readability.

## TECHNICAL REQUIREMENTS
- **Framework**: React with TypeScript to ensure type safety and component-based architecture.
- **Styling**: Tailwind CSS for utility-first styling, allowing rapid design changes and responsiveness.
- **UI Components**: Utilize shadcn/ui for pre-built components that fit the minimalist aesthetic.
- **State Management**: Implement React Query or Zustand for efficient state management of live data.

## IMPLEMENTATION STEPS
1. **Set Up Project**: Initialize a new React project with TypeScript and configure Tailwind CSS.
2. **Audio Capture Integration**: Use WebRTC or a similar API to capture and stream live audio.
3. **Speech-to-Text Integration**: Integrate a streaming speech-to-text API (e.g., Whisper or Deepgram) to convert audio to text in real-time.
4. **NLP Model Development**: Develop the NLP model for scam detection, utilizing both rule-based logic and AI reasoning to analyze transcripts.
5. **Dynamic Risk Scoring**: Create a component that calculates and displays the fraud risk score based on ongoing analysis.
6. **Alert System**: Implement a notification system that triggers instant alerts when the risk score exceeds the defined threshold.
7. **UI Development**: Build out the user interface components, ensuring responsiveness and a consistent minimalist style throughout.

## USER EXPERIENCE
Key user interactions include initiating a call, receiving live updates on the transcript, viewing the dynamic risk score, and responding to alerts. Users will appreciate the intuitive layout that allows for easy monitoring of their calls without distractions, ensuring a focus on communication while being safeguarded against potential fraud. The alerts will be designed to be noticeable but not disruptive, providing a balance between user experience and security.