Tag
4 posts
whisper.cpp is the C and C++ port of OpenAI Whisper that runs fully offline, with model sizes from 75 MB to 2.9 GB and accuracy that scales with size. This guide covers install, batch transcription of voice memos, the limits, and where a managed tool fills the gaps.
A practical how-to for turning speech into markdown. Covers the transcript-to-structure pipeline, a copy-paste output example, Obsidian and Git workflows, quik.md's built-in flow, and an honest look at DIY Whisper setups.
Voice-to-text for notes crossed the usable threshold around 2023. This guide covers what you should expect from accuracy, where it still fails, and the apps worth using in 2026 for knowledge work.
Voice-to-task capture turns a thought into a sorted todo in seconds. This guide walks through the mechanics, the accuracy numbers, and the failure modes that decide whether the workflow actually sticks.
