A real-time speech transcription application optimized for Indian English accents, similar to Otter.ai. Users can join channels where their speech is transcribed in real-time and visible to all other users in the channel.

• Real-time speech-to-text transcription optimized for Indian English accents
• Channel-based communication where multiple users can join
• Live updates as users speak or type
• Mobile-friendly interface
• User authentication and profile management

├── backend/
│   ├── database/
│   │   ├── migrations.ts    # Schema definitions
│   │   └── queries.ts       # DB query functions
│   ├── routes/              # Route modules
│   │   ├── auth.ts          # Authentication routes
│   │   ├── channels.ts      # Channel management
│   │   └── static.ts        # Static file serving
│   ├── websocket/
│   │   └── handler.ts       # WebSocket connection handling
│   └── index.ts             # Main entry point
├── frontend/
│   ├── components/
│   │   ├── App.tsx
│   │   ├── ChannelView.tsx
│   │   ├── Login.tsx
│   │   └── Transcription.tsx
│   ├── favicon.svg
│   ├── index.html           # Main HTML template
│   ├── index.tsx            # Frontend JS entry point
│   └── style.css
└── shared/
    ├── types.ts             # Shared type definitions
    └── utils.ts             # Shared utility functions

• Backend: Hono.js with WebSockets
• Frontend: React with TypeScript
• Database: SQLite for persistent storage
• Speech Recognition: Web Speech API with customization for Indian English
• Styling: TailwindCSS

1. Users create an account or log in
2. They can create a new channel or join an existing one
3. Once in a channel, they can start speaking or typing
4. The speech is transcribed in real-time using the Web Speech API
5. Transcriptions are sent to all users in the channel via WebSockets
6. All users see the transcriptions in real-time with speaker identification

• Speech recognition quality depends on the browser's implementation of Web Speech API
• Optimization for Indian English accents is achieved through configuration but may not be perfect