stt

stt is a Claude Code plugin that turns spoken recordings into Markdown notes you can drop into a knowledge vault such as Obsidian. It has one skill per source: /stt:clovanote imports transcripts and speaker labels that Naver Clova Note has already produced, and /stt:apple transcribes Apple Voice Memos or exported audio files locally with whisper.cpp. Apple transcription never uploads the recording. Installing the plugin does not install the transcription engine or model, so ffmpeg, whisper-cli and a GGML model have to be set up separately. Clova Note has no public API; the plugin replays the web app's internal API with a logged-in session, and the first login needs a person to clear Naver's protection step once. Transcripts carry timestamps but no speaker separation or summary, and Whisper can mishear or hallucinate on silence, so important passages should be checked against the audio.