Help us improve
Share bugs, ideas, or general feedback.
Cleans raw interview transcripts into polished Q&A format by removing filler words, fixing speaker labels, standardising punctuation, and optionally adding timestamps.
npx claudepluginhub ur-grue/autopunk-media-skills --plugin autopunk-media-skillsHow this skill is triggered — by the user, by Claude, or both
Slash command
/autopunk-media-skills:transcript-qa-formatterThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Cleans a raw interview transcript into a polished Q&A — removing filler words, fixing speaker labels, standardising punctuation, and optionally inserting timestamps — ready to publish or hand to a copy editor.
Cleans raw auto-generated podcast transcripts for publication: removes filler words, corrects errors, adds speaker labels, and formats for readability while preserving authentic voice.
Mechanically cleans raw dictation transcripts into editable drafts: removes fillers/false starts, restores punctuation/paragraphs, flags transcription errors without editing content or voice.
Converts raw meeting transcript .txt files into structured .md notes with metadata, TL;DR, key topics, action items, and quotes. Useful for processing meeting recordings or chat logs.
Share bugs, ideas, or general feedback.
Cleans a raw interview transcript into a polished Q&A — removing filler words, fixing speaker labels, standardising punctuation, and optionally inserting timestamps — ready to publish or hand to a copy editor.
Required: The raw transcript text, pasted in full. Include speaker labels if they exist, even if inconsistent. Optional: Real names to replace placeholder labels (e.g. "Speaker 1 = Dr. Amara Nwosu"); preferred timestamp style (HH:MM:SS or MM:SS); whether you want light cleaning (filler words only) or full editing (also tighten run-on answers); the publication's house style for Q&A layout (e.g. bold Q, italic A, or labelled by name).
Full cleaned transcript in Q&A layout. Length mirrors the source — no content is cut beyond filler words. Tone follows the speaker's own register (formal interview stays formal; conversational stays conversational). A brief change log follows the transcript: number of filler words removed, any speaker labels corrected, any ambiguous turns flagged for editor review. No commentary or suggestions beyond what was asked.
Speaker 1: So, um, the, uh, the main challenge we faced — and I think this is, you know, sort of the crux of it — was that the supply chain data just wasn't, like, granular enough for, um, regional analysis. We ended up, uh, building our own tracking system from scratch.
UNKNOWN: And how long did that take?
Speaker 1: About, um, fourteen months. Which is, you know, not ideal when you're working to a quarterly reporting cycle. But we felt it was the right call.
UNKNOWN: Were there other options you considered?
Speaker 1: Yeah, we looked at, uh, three off-the-shelf platforms. None of them had the, you know, the regional breakdown we needed. So, um, it was really a build-versus-buy decision and, I mean, build won.
Speaker names: Speaker 1 = Helena Varga (Head of Operations); UNKNOWN = Interviewer
Timestamp style: MM:SS at each speaker turn
Layout: Bold Q, plain A, labelled by role
**Q [00:00]:** And how long did that take?
A [00:08] Helena Varga (Head of Operations): About fourteen months. Which is not ideal when you're working to a quarterly reporting cycle. But we felt it was the right call.
**Q [00:18]:** Were there other options you considered?
A [00:21] Helena Varga (Head of Operations): Yeah, we looked at three off-the-shelf platforms. None of them had the regional breakdown we needed. So it was really a build-versus-buy decision and build won.
Note: The opening speaker turn by Helena Varga appears before the first interviewer question in the source. It has been placed at the top as a standalone answer block for context:
A [00:00] Helena Varga (Head of Operations): The main challenge we faced — and I think this is the crux of it — was that the supply chain data wasn't granular enough for regional analysis. We ended up building our own tracking system from scratch.
Change log: