PeresteParse

A privacy-focused macOS study application that transforms voice recordings into Anki flashcards using locally-run large language models.

Overview

PeresteParse combines voice recording, speech-to-text transcription, and LLM-powered parsing to streamline the creation of study flashcards. All processing happens locally on your device, ensuring complete privacy of your study materials.

Key Features

• Voice recording with Parakeet MLX speech-to-text transcription
• LLM-powered flashcard generation (Claude API or local Qwen3-4B)
• Export to Anki (.apkg) or CSV formats
• Session tracking and statistics
• Local data storage for complete privacy
• Automatic recording duration tracking

System Requirements

• macOS 11.0 or later
• 4GB+ RAM (8GB+ recommended for local LLM)
• 5GB available storage

Technical Details

Built with SwiftUI for a native macOS experience, PeresteParse leverages state-of-the-art machine learning models for speech recognition and natural language processing. The app supports both cloud-based (Claude API) and fully local (Qwen3-4B) LLM processing, giving you flexibility based on your privacy preferences and internet connectivity.

View source code on GitHub