what's in the export
- filler removed — "um," "uh," "like," "you know," "i mean," and the half-finished restarts speakers do mid-sentence. removing these is conventional in journalistic quote handling and most style guides explicitly allow it.
- obvious disfluencies fixed — false starts ("the — the model says…"), repeated words ("it it doesn't work"), and clear vocal stumbles. the edit is conservative: if there's any doubt about whether something was a stumble or intentional, the tool leaves it alone and flags it for your decision.
- exact wording preserved — no paraphrasing, no synonym substitution, no "smoothing." the words inside quote marks are the words the speaker actually said. you decide what counts as verbatim, not the tool.
- timestamp anchor on every quote so you can verify against the audio before publication. in the .docx export, each quote carries a footnote with its timestamp:
"the model is the moat."¹with¹ [00:21:48]in the footer. click in the editor to play that second. - attribution preserved exactly as it appeared in the audio — "X said," "according to X," "X told me." you decide which framing to use in the published piece; the export gives you all of them.
- elision marked, not hidden — if the cleanup pass ever removes more than a filler word from a quote, the elision is marked with
[…]in the standard journalistic convention. you see exactly what was cut. the default is conservative: most quotes export with no elisions at all. - .docx with footnoted timestamps for the standard newsroom workflow. also exports as markdown if you draft in markdown and convert later.
sample output
a 75-minute recorded interview, with eight quotes marked in the editor, exports as a clean-quotes.docx roughly like this:
# quotes — interview with senator alvarez, 2026-04-22
1. "we knew the rule was unenforceable the day we passed it."¹
— sen. alvarez
2. "every committee member was told the projections, and every
committee member voted yes anyway."²
— sen. alvarez
3. "i don't regret the vote. i regret the explanation we gave
afterward."³
— sen. alvarez
4. "if you ask me what i would change, it would be the timeline,
not the policy."⁴
— sen. alvarez
---
¹ [00:08:14] — audio: alvarez-2026-04-22.m4a
² [00:14:33] — audio: alvarez-2026-04-22.m4a
³ [00:21:08] — audio: alvarez-2026-04-22.m4a
⁴ [00:42:55] — audio: alvarez-2026-04-22.m4a
the timestamps in the footer are the verification trail. click any quote in the editor before export to confirm the wording against the audio. for fact-checking and legal review, the .docx travels alongside the audio file with the timestamps as the index.
the workflow
- transcribe the interview. upload the recording — phone voice memo, zoom recording, riverside, or a digital recorder. a 75-minute interview is ready in about three minutes.
- read through and mark quotes in the editor. highlight any sentence and tag it as a quote. the timestamp anchor is captured automatically. you can mark partial sentences, full paragraphs, or exchanges between speakers.
- review the cleanup diff per quote. for each marked quote, the editor shows the raw audio transcription side-by-side with the cleaned version. accept or reject the cleanup individually. the default cleanup is filler-removal only.
- verify against audio. click any word to play that second. for any quote going into print, listen end-to-end before export. the editor keeps a "verified" flag per quote so you can track which quotes you've checked.
- export clean-quotes.docx. with footnoted timestamps. the file is ready for the copy desk and for the fact-checker, with the audio file referenced by name in every footnote.
when to use this vs. other exports
- clean-quotes export (this page) — for journalism, fact-checking, and any workflow where quote accuracy is the deliverable and verification is non-negotiable.
- blog post draft — when the deliverable is a long-form prose piece rather than a set of verified quotes. heavier on narrative structure, lighter on per-quote verification.
- show-notes export — for podcast episode pages with chapter timestamps and pull-quotes presented as blockquotes.
- raw transcript — for legal review, court filings, and any workflow that requires every word with timestamps on every line and no cleanup.
privacy
for source-protected reporting, embargoed interviews, investigative work, and any audio that can't leave the newsroom, run the file in private mode. the clean-quotes export works identically; the audio, transcript, and verification trail stay on your laptop. this is the default mode for most journalism workflows.
pricing for clean-quotes export
$0.25 per minute, all exports included. the same recording gives you the raw transcript, the clean-quotes .docx, and any other format you need from the same audio without an additional charge. no per-format upcharge, no subscription, no minimum. waitlist signups get the first month free and 50% off forever.