Speech to Markdown with Local Models (github.com)

A new tool, Speech-to-Markdown, has been launched, enabling users to convert their spoken words into structured Markdown documents entirely through local applications on macOS and iOS. The desktop app utilizes the whisper.cpp engine for speech-to-text processing and interacts with any local OpenAI-compatible server, while the mobile app employs Apple's SpeechAnalyzer and Foundation Models, ensuring complete offline functionality without the need for cloud services or API integration. This development is significant for the AI/ML community as it emphasizes privacy and local processing, a growing demand among users wary of data leaks. It introduces a streamlined user experience with features such as real-time transcription and editing modes—allowing users to dictate, edit, or append text without latency issues—even as document size increases. The integration of local models and zero setup requirements makes it an accessible tool for developers and everyday users alike, showcasing the potential of on-device processing powered by advanced local language models.