23: Transcription and the Listening Companion - transcript
Liam: The QUILL Cast, episode twenty-three. I'm Liam. Today: transcription and the Listening Companion, the pipeline that turns recordings into finished documents.
Jessica: I'm Jessica. The pipeline has three stages, hold the shape: capture, any audio or video file, meetings, lectures, interviews, voice memos. Transcribe, on your machine, privately. Transform, from raw transcript to the document you actually needed.
Liam: Stage two first, with the respect it deserves. Local transcription means the two-hour recording of a confidential board meeting is processed entirely on your own processor. For anyone touching client calls, medical conversations, legal matters, student records, that fact alone is the feature. Options: translate to English during transcription, or identify different speakers so the transcript shows who said what.
Jessica: Point QUILL at the file from the speech tools, go do something else, background work never blocks the editor, and return to a transcript document in a tab. Any format your recorder produces, more or less, and if the audio-format helper ffmpeg is installed, from Download Optional Components, the coverage gets even wider.
Liam: Now stage three, where the Companion earns its name, because nobody actually wants a transcript. A transcript is a wall of spoken language, filler, tangents, no structure. People want what's in it. Enter Transcript Actions.
Jessica: The built-in menu: Meeting Minutes, structured with topics. Action Items, just the commitments, who and what. Executive Summary. Interview Notes, and Study Notes for lectures. Q and A extraction. Key Quotes, verbatim. A Decisions Log. Follow-Up Email, drafted ready to send. And Clean Draft, which tidies language while keeping substance. Run any of them on the open transcript, and, quietly powerful, on any document at all, Key Quotes on a long report is excellent.
Liam: These actions run through your AI connection, part five's topic, with the standard contract: results arrive as reviewable documents, nothing silent. And if AI isn't configured yet, transcription itself still works fully, it's local machinery, the transform tier is the optional layer.
Jessica: The customization tier: the Action Builder. The built-ins cover common cases; the Builder covers yours, as a no-syntax form: name the action, describe what it should produce, set the structure. A support lead builds Complaint Summary, issue, temperature, promised follow-ups. A student builds Flashcard Prep, lecture to question-answer pairs. Your actions join the same menu as the built-ins, and the tool stops being generic.
Liam: And the automation crown from episode twenty: attach transcription plus a transcript action to a watch folder, and the pipeline runs itself. Recorder syncs to folder, minutes await you. The committee secretary's whole job shifts from typing while listening to reviewing a draft, which is the part that actually needs a human.
Jessica: Craft notes. Microphones again, a cheap conference mic improves transcripts more than any setting. Name recordings meaningfully before the watch folder eats them, the filename rides the whole pipeline. And for long recordings, run Executive Summary first, read it, then decide which deeper actions the recording deserves, ten seconds of summary can save processing ninety minutes of the wrong file.
Liam: Homework. One: transcribe any recording you have, even a voice memo, just to see the pipeline run. Two: run two different actions on the same transcript and compare what each kept. Three: sketch, on paper is fine, the custom action you'd build for a document you produce monthly, next episode block you'll get to build it properly.
Jessica: Next episode: text back out to audio, speech exports, chaptered audiobooks, and the DAISY talking books format for accessible libraries.
Liam: I'm Liam.
Jessica: I'm Jessica. Stop taking minutes; start taking review passes.