Zesty - Voice-Assisted Culinary Companion

A sophisticated Android cooking application that combines offline recipe management with voice-assisted cooking features. Zesty provides hands-free cooking guidance through advanced machine learning-powered voice recognition and offers access to over 6000+ Indian recipes offline.

🎬 Demo Video

Watch Zesty in action - Voice-assisted cooking with 6000+ offline recipes

📺 Detailed Walkthrough

For a comprehensive walkthrough of all features and voice-assisted cooking capabilities:

🎥 Complete App Walkthrough - Detailed demonstration of voice commands, recipe search, and cooking mode

▶️ Watch on YouTube

📱 Application Screenshots

📱 View All Screenshots

Click to see detailed screenshots of all app screens

📱 Features

🎯 Core Features

Voice-Assisted Cooking: Hands-free recipe navigation using ML-powered voice commands
Offline Recipe Database: 6000+ Indian recipes accessible without internet
Smart Recipe Search: Advanced fuzzy search with ingredient-based filtering
Recipe Management: Favorites, shopping lists, and personal ingredient tracking
Cooking Timer: Voice-controlled timer with step-by-step guidance
YouTube Integration: Video tutorials for recipes
Modern UI: Beautiful Material Design 3 interface with smooth animations

🧠 AI-Powered Features

Intent Classification: ML model trained to understand cooking voice commands
Speech Processing: Real-time voice command processing and execution
Smart Search: Fuzzy matching and ingredient-based recipe discovery
Text-to-Speech: Audio feedback for cooking instructions

🏗️ Architecture Overview

Application Flow Diagram

graph TB
    A[User Opens App] --> B[Splash Screen]
    B --> C[Home Screen]
    C --> D{User Action}
    
    D -->|Voice Command| E[Speech Recognition]
    D -->|Text Search| F[Search Input]
    D -->|Browse| G[Recipe Categories]
    
    E --> H[Intent Classification]
    H --> I{Command Type}
    
    I -->|Search Recipe| J[Local Recipe Search]
    I -->|Timer Command| K[Cooking Timer]
    I -->|Navigation| L[Step Navigation]
    I -->|Show Ingredients| M[Ingredient Display]
    
    J --> N{Local Results?}
    N -->|Yes| O[Display Local Recipes]
    N -->|No| P[TheMealDB API]
    P --> Q[Display Online Recipes]
    
    F --> J
    G --> J
    
    O --> R[Recipe Detail Screen]
    Q --> R
    
    R --> S[Cooking Mode]
    S --> T[Voice-Guided Steps]
    T --> U[Timer Management]
    U --> V[Step Navigation]
    V --> T
    
    R --> W[Add to Favorites]
    R --> X[Add to Shopping List]
    R --> Y[Share Recipe]

Data Flow Architecture

graph LR
    A[User Voice Input] --> B[SpeechHelper]
    B --> C[IntentClassifier]
    C --> D[SpeechProcessor]
    D --> E[Command Execution]
    
    F[Recipe Search] --> G[LocalRecipeHelper]
    G --> H[Offline Database]
    G --> I[Fuzzy Search Engine]
    
    J[Online Fallback] --> K[TheMealDB API]
    K --> L[Recipe Conversion]
    L --> M[Display Results]
    
    N[User Actions] --> O[Room Database]
    O --> P[Favorites]
    O --> Q[Shopping List]
    O --> R[User Ingredients]

📁 Project Structure

🎯 Core Application Files

Main Application

MainActivity.kt - Entry point with global error handling
ZestyApplication.kt - Application class with dependency injection
AndroidManifest.xml - Permissions and app configuration

Navigation & UI

NavGraph.kt - Navigation structure with bottom bar
HomeScreen.kt - Main dashboard with voice search
RecipeDetailScreen.kt - Detailed recipe view with cooking mode
FavoritesScreen.kt - Saved recipes management
ShoppingListScreen.kt - Shopping list functionality
MyIngredientsScreen.kt - Personal ingredient tracking & detailed ingredient search

Data Layer

Models.kt - Core data models (Recipe, Ingredient, etc.)
ZestyDatabase.kt - Room database configuration
ZestyDao.kt - Database access methods
FavoritesRepository.kt - Repository pattern implementation

ViewModels

HomeViewModel.kt - Home screen state management
RecipeViewModel.kt - Recipe detail and cooking state
FavoritesViewModel.kt - Favorites management
ShoppingListViewModel.kt - Shopping list operations
MyIngredientsViewModel.kt - Ingredient management

🤖 AI & Voice Processing

Voice Recognition System

SpeechHelper.kt - Android Speech Recognition integration
SpeechProcessor.kt - Voice command processing and intent extraction
IntentClassifier.kt - ML model for command classification
TTSHelper.kt - Text-to-speech functionality

Machine Learning Model

model.tflite - Trained TensorFlow Lite model for intent classification
labels.txt - Command labels for the ML model
tflite_voice_training_model.ipynb - Jupyter notebook for model training

🍽️ Recipe Management

Offline Recipe System

LocalRecipeHelper.kt - Offline recipe search and management
recipes.json - 6000+ Indian recipes database

Online Recipe Integration

TheMealDBHelper.kt - TheMealDB API integration
TheMealDBModels.kt - API response models
NetworkHelper.kt - Network connectivity management

🛠️ Utilities & Helpers

Core Utilities

Constants.kt - Application constants
Extensions.kt - Kotlin extension functions
RecipeRepository.kt - In-memory recipe storage
MicButton.kt - Custom microphone button component

Specialized Helpers

ImageCacheHelper.kt - Image loading and caching
RecipeStepParser.kt - Recipe instruction parsing
SearchPerformanceHelper.kt - Search optimization
TimerWorker.kt - Background timer management
YouTubeAPIHelper.kt - YouTube video integration

🧠 Machine Learning Implementation

Voice Command Classification

The application uses a custom-trained TensorFlow Lite model for voice command classification:

Model Architecture

Input: Text strings (voice commands)
Architecture: Bidirectional LSTM with Global Max Pooling
Output: Intent classification (5 main intents)
Accuracy: 95%+ on test data

Supported Voice Commands

search_recipe    - "I want to make biryani"
set_timer        - "Set timer for 10 minutes"
show_ingredients - "Show me the ingredients"
next_step        - "Next step"
previous_step    - "Go back"

Training Process

The model was trained using Google Colab with the following approach:

Data Collection: 794 voice command samples across 5 intent categories
Preprocessing: Text normalization and tokenization
Model Training: BiLSTM with embedding layer and dropout
Optimization: Early stopping and learning rate reduction
Export: TensorFlow Lite conversion for mobile deployment

Model Performance

Training Accuracy: 95%+
Validation Accuracy: 93%+
Model Size: ~2MB (optimized for mobile)
Inference Time: <100ms on average Android device

🍽️ Recipe Database System

Offline Recipe Management

Zesty includes a comprehensive offline recipe database with 6000+ Indian recipes:

Dataset Source

Original Dataset: Kaggle Indian Recipes Dataset
Processing: Custom Python scripts for data cleaning and image URL updates
Format: JSON with structured recipe data

Data Processing Pipeline

graph LR
    A[Kaggle Dataset] --> B[python_image_gen.py]
    B --> C[Updated CSV]
    C --> D[python_to_json.py]
    D --> E[recipes.json]
    E --> F[Android Assets]
    F --> G[LocalRecipeHelper]

Recipe Data Structure

{
    "RecipeName": "masala karela recipe",
    "TotalTimeInMins": 45,
    "Instructions": ["Step 1", "Step 2", ...],
    "Ingredients": ["salt", "onion", ...],
    "image-url": "https://...",
    "Ingredient-count": 10,
    "matched_page_url": "https://..."
}

Search Capabilities

Fuzzy Search: Levenshtein distance-based matching
Ingredient Matching: Find recipes by available ingredients
Title Search: Exact and partial title matching
Performance: Sub-100ms search times with caching

Online Recipe Fallback

When offline recipes don't match user queries, the app falls back to TheMealDB API:

API Integration

Base URL: https://www.themealdb.com/
Endpoints: Search, random recipes, recipe details
Fallback Strategy: Local first, then online
Caching: Search results cached for performance

🎤 Voice-Assisted Cooking Flow

Voice Command Processing Pipeline

sequenceDiagram
    participant U as User
    participant SH as SpeechHelper
    participant IC as IntentClassifier
    participant SP as SpeechProcessor
    participant VM as ViewModel
    participant TTS as TTSHelper
    
    U->>SH: Voice Command
    SH->>IC: Text Input
    IC->>SP: Intent Classification
    SP->>VM: Processed Command
    VM->>TTS: Audio Feedback
    TTS->>U: Spoken Response

Cooking Mode Features

Step-by-Step Guidance

Voice-controlled navigation between recipe steps
Audio feedback for each cooking instruction
Timer integration for timed cooking steps
Hands-free operation during cooking

Timer Management

Voice-activated timer setting
Background timer with notifications
Step-specific timing recommendations
Timer pause/resume functionality

🛠️ Technical Implementation

Architecture Patterns

MVVM Architecture

Model: Data classes and Room entities
View: Jetpack Compose UI components
ViewModel: State management with StateFlow
Repository: Data access abstraction

Dependency Injection

Manual dependency injection through Application class
Repository pattern for data access
Factory pattern for ViewModel creation

Key Technologies

Android Technologies

UI: Jetpack Compose with Material Design 3
Navigation: Navigation Compose
Database: Room with TypeConverters
Networking: Retrofit with Gson
Background Tasks: WorkManager
Image Loading: Coil with GIF support

Machine Learning

Framework: TensorFlow Lite
Model: Custom BiLSTM (Bidirectional Long Short Memory) for intent classification
Delegates: Flex Delegate for enhanced operations
Performance: Optimized for mobile inference

Voice Processing

Recognition: Android Speech Recognition API
TTS: Android Text-to-Speech API
Processing: Custom intent extraction algorithms

Performance Optimizations

Search Performance

Caching: Multi-level search result caching
Indexing: Pre-built search indices for fast lookups
Fuzzy Matching: Optimized Levenshtein distance implementation
Background Processing: Async search operations

Memory Management

Image Caching: Efficient image loading with Coil
Database Optimization: Room with proper indexing
State Management: Efficient StateFlow usage
Resource Cleanup: Proper lifecycle management

📱 User Interface Design

Design System

Material Design 3

Color Scheme: Custom color palette with dynamic theming
Typography: Material Design typography scale
Components: Custom components with consistent styling
Animations: Smooth transitions and micro-interactions

Key UI Components

MicButton: Custom animated microphone button
CollapsibleHeader: Dynamic header with search functionality
RecipeCard: Consistent recipe display cards
CustomNavigationBar: Animated bottom navigation

Screen Layouts

Home Screen

Featured recipe carousel
Voice search interface
Suggested recipes grid
Quick access to favorites

Recipe Detail Screen

Full-screen recipe images
Step-by-step cooking instructions
Voice control interface
Timer and navigation controls

Cooking Mode

Hands-free operation
Voice feedback
Timer integration
Step navigation

🔧 Setup & Installation

Prerequisites

Android Studio Arctic Fox or later
Android SDK 26+ (Android 8.0+)
Kotlin 2.0.0+
Gradle 8.4.2+

Installation Steps

Clone the Repository

git clone https://github.com/indra057/zesty.git
cd zesty

Configure API Keys
- Create local.properties file in project root
- Add your YouTube API key:
```
YOUTUBE_API_KEY=your_youtube_api_key_here
```
Build and Run
```
./gradlew assembleDebug
```

Required Permissions

The app requires the following permissions:

RECORD_AUDIO - For voice recognition
INTERNET - For online recipe fetching
ACCESS_NETWORK_STATE - For network status checking
POST_NOTIFICATIONS - For timer notifications

🧪 Testing

Test Coverage

Unit Tests: ViewModel and utility function testing
Integration Tests: Database and API integration testing
UI Tests: Compose UI component testing
Voice Tests: Speech recognition accuracy testing

Running Tests

# Unit tests
./gradlew test

# Instrumented tests
./gradlew connectedAndroidTest

# All tests
./gradlew check

🚀 Performance Metrics

App Performance

Cold Start: <2 seconds
Search Response: <100ms average
Voice Recognition: <500ms average
Memory Usage: <150MB typical
Battery Impact: Minimal with optimized background tasks

ML Model Performance

Model Size: 2MB
Inference Time: <100ms
Accuracy: 95%+ on test data
Memory Usage: <10MB during inference

🔮 Future Enhancements

Planned Features

Multi-language Support: Hindi and regional language support
Recipe Recommendations: AI-powered recipe suggestions
Nutritional Analysis: Calorie and nutrition tracking
Social Features: Recipe sharing and community
Offline Video: Cached cooking videos
Smart Shopping: AI-powered shopping list optimization

Technical Improvements

Model Updates: Continuous ML model improvements
Performance: Further optimization for low-end devices
Accessibility: Enhanced accessibility features
Testing: Expanded test coverage
Documentation: API documentation and developer guides

🤝 Contributing

We welcome contributions! Please see our Contributing Guidelines for details.

Development Setup

Fork the repository
Create a feature branch
Make your changes
Add tests for new functionality
Submit a pull request

Code Style

Follow Kotlin coding conventions
Use meaningful variable and function names
Add comments for complex logic
Maintain consistent formatting

📄 License

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

🙏 Acknowledgments

Dataset: Kaggle Indian Recipes Dataset
API: TheMealDB for additional recipes
ML Framework: TensorFlow Lite for mobile machine learning
UI Framework: Jetpack Compose for modern Android UI

📞 Support

For support, questions, or feature requests:

Issues: GitHub Issues
Discussions: GitHub Discussions
Email: indrajiitdas057@gmail.com

📱 Application Screenshots {#application-screenshots}

Experience the complete Zesty cooking journey through our intuitive interface:

Splash

Home

Search

Results

Ingredients

Recipe Detail

Cooking Mode

Timer

Favorites

Shopping List

Made with ❤️ for cooking enthusiasts

Zesty - Your Culinary companion

Name		Name	Last commit message	Last commit date
Latest commit History 8 Commits
.idea		.idea
.vscode		.vscode
app		app
gradle		gradle
.gitignore		.gitignore
README.md		README.md
build.gradle.kts		build.gradle.kts
gradle.properties		gradle.properties
gradlew		gradlew
gradlew.bat		gradlew.bat
settings.gradle.kts		settings.gradle.kts

indra057/Zesty

Folders and files

Latest commit

History

Repository files navigation