NovaTV 🎙️

A modern voice-powered AI assistant for Android TV, built with Jetpack Compose for TV.

- Screenshot:

- Recording:

Screen_recording_20251207_204854.mp4

Note: If the video doesn't play above, you can find it at readme_resources/Screen_recording_20251207_204854.mp4

✨ Features

• 🎤 Voice Input — Press the mic button to speak your questions
• 🤖 AI-Powered Responses — Powered by Cloudflare Workers AI (Llama 3.1 8B)
• 🔊 Text-to-Speech — Bot responses are read aloud automatically
• 📺 TV-First Design — Built specifically for Android TV with D-pad navigation
• 🎨 Modern Dark UI — Sleek indigo/purple accent theme

🏗️ Architecture

Voice Flow

🎤 Record Audio ──► 📝 Speech-to-Text ──► 🤖 LLM ──► 🔊 Text-to-Speech
     (Mic)              (Whisper)        (Llama)      (Android TTS)

Layer Diagram

┌──────────────────────────────────────────────────────────────┐
│                        UI Layer                              │
│  ┌─────────────┐  ┌───────────────┐  ┌────────────────────┐  │
│  │ HomeScreen  │  │ HeaderWidget  │  │ VocalAssistantVM   │  │
│  │ (Compose)   │  │ (Compose)     │  │ (STT + LLM + TTS)  │  │
│  └─────────────┘  └───────────────┘  └────────────────────┘  │
└──────────────────────────────────────────────────────────────┘
                              ▲
                              │
                              ▼
┌──────────────────────────────────────────────────────────────┐
│                      Domain Layer                            │
│  ┌─────────────────────┐  ┌────────────────────────────────┐ │
│  │ SpeechToTextUseCase │  │ TextGenerationUseCase          │ │
│  └─────────────────────┘  └────────────────────────────────┘ │
└──────────────────────────────────────────────────────────────┘
                              ▲
                              │
                              ▼
┌──────────────────────────────────────────────────────────────┐
│                       Data Layer                             │
│  ┌─────────────────────────────────────────────────────────┐ │
│  │ CloudflareRepositoryImpl                                │ │
│  │  • Speech-to-Text (Whisper)                             │ │
│  │  • Text Generation (Llama 3.1 8B Instruct)              │ │
│  └─────────────────────────────────────────────────────────┘ │
└──────────────────────────────────────────────────────────────┘

🚀 Getting Started

Prerequisites

• Android Studio Hedgehog or newer
• Android TV device or emulator (API 21+)
• Cloudflare account with Workers AI access

Setup

1. Clone the repository

   git clone https://github.com/yourusername/ChatbotTVCompose.git
   cd ChatbotTVCompose

2. Configure API credentials

   Copy the example environment file:

   cp .env.example .env

   Edit .env and add your Cloudflare credentials:

   CLOUDFLARE_ACCOUNT_ID=your_account_id
   CLOUDFLARE_API_TOKEN=your_api_token

3. Build and run

   ./gradlew :tv:installDebug

🎮 Usage

Action	Control
Navigate	D-pad arrows
Select/Confirm	Enter / Center button
Start recording	Select mic button, speak
Stop recording	Select mic button again

Header Buttons

Button	Function
🔔 Notifications	Open app notification settings
📱 Apps	Open system apps manager
⚙️ Settings	Open Android settings
🎤 Mic	Start/stop voice recording

🛠️ Tech Stack

Category	Technology
UI	Jetpack Compose for TV, Material 3
Architecture	MVVM, Clean Architecture
DI	Hilt
Networking	Retrofit, OkHttp
Serialization	Kotlinx Serialization
AI Provider	Cloudflare Workers AI
STT Model	OpenAI Whisper
LLM Model	Meta Llama 3.1 8B Instruct (Fast)
TTS	Android TextToSpeech

📁 Project Structure

tv/src/main/java/com/applicassion/ChatbotTVCompose/
├── data/
│   ├── remote/
│   │   └── ai_provider/
│   │       ├── BaseAIService.kt          # Generic AI service interface
│   │       └── cloudflare/
│   │           ├── CloudflareWorkersBaseAIService.kt
│   │           └── dto/                   # Data Transfer Objects
│   └── repository_impl/
│       └── CloudflareRepositoryImpl.kt
├── di/
│   └── CloudflareModule.kt               # Hilt dependency injection
├── domain/
│   ├── model/                            # Domain models
│   ├── repository/
│   │   └── AIRepository.kt
│   └── usecase/
│       ├── SpeechToTextUseCase.kt
│       └── TextGenerationUseCase.kt
├── ui/
│   ├── screens/
│   │   └── home/
│   │       └── HomeScreen.kt
│   ├── widgets/
│   │   ├── HeaderWidget.kt
│   │   ├── ChatBubble.kt
│   │   └── TvCircularProgressIndicator.kt
│   ├── theme/
│   │   └── Color.kt
│   └── VocalAssistantViewModel.kt
└── utils/
    └── Constants.kt

🔐 Environment Variables

Variable	Description
CLOUDFLARE_ACCOUNT_ID	Your Cloudflare account ID
CLOUDFLARE_API_TOKEN	API token with Workers AI permissions

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Cloudflare Workers AI for the AI models
• Jetpack Compose for TV for the UI framework
• Material Design 3 for design guidelines

---

Built with ❤️ for Android TV