Voice memo transcripts

Each evening, transcribe the new audio in my workspace and file it as clean notes.

Copy

What it does

Drop audio into your workspace during the day. In the evening the agent transcribes whatever is new, cleans the text up into a readable note — paragraphs, no filler — and files it in your documents next to everything else you write.

Once it is a note it is searchable and the agent can work with it: summarize it, pull the tasks out of it, quote it in something else.

Setup

An OpenAI API key, set on the Whisper skill’s card in Skills. Files go up to 25 MB each; longer recordings need splitting first, which the ffmpeg toolkit in the Media app can do.

Transcripts are written into documents/ in your workspace. No channel to connect.

Installing it

Install it from Skills, in the routines shelf. It arrives as a draft task with the schedule above already seeded and its trigger disarmed, so nothing runs until you arm it. The agent then fills the blanks the template leaves — which folder to watch, what the notes should be named, which language — with you, in chat.

The schedule and the delivery channel are yours to change at install time and after: both are plain fields on the task.