Sweeft AI Voice-to-JSON is an advanced Proof of Concept (POC) developed for iOS 26 that enables users to interact with business intelligence data using natural language voice commands.
The application captures user speech, converts it into text using Apple's latest Speech Framework APIs, understands business intent through on-device AI processing, and generates structured JSON queries that can be consumed by analytics and reporting systems.
The solution supports multilingual voice interaction (English and French), automatic speech detection, business entity extraction, schema validation, and JSON query generation while ensuring complete on-device processing for privacy and performance.
Key Features:
Business Use Cases
Technology Stack
Mobile Development: Swift, SwiftUI, iOS 26.
Apple Speech Framework: SpeechAnalyzer, SpeechTranscriber, SpeechDetector, AssetInventory.
Artificial Intelligence: Foundation Models Framework, SystemLanguageModel, LanguageModelSession, On-Device AI Inference.
Audio Processing: AVFoundation, AVAudioEngine, AVAudioSession, AVAudioConverter.
Architecture: MVVM, Combine, Async/Await, AsyncStream.
Performance & Profiling: Xcode Instruments, CPU Monitoring, Memory Profiling, Energy Impact Analysis, Performance Benchmarking.
Project Highlights