A fact-checker asking for the exact wording, three weeks later.

The moment a recorded interview earns its keep isn't during the call — it's weeks later, when an editor or a fact-checker asks whether a source really said "considering" or "planning," and the answer has to come from the recording rather than a reporter's memory of how the sentence probably went. A notebook gets the gist. A recording, actually transcribed, gets the exact words a correction later depends on.

What a bot-based tool solves, and what it changes about the room

Otter joins a video interview as a named participant, which works cleanly for a scheduled call on a platform that allows it, and its own site lists mobile apps for iOS and Android that let a reporter record straight from a phone — a real advantage on a story that starts with somebody calling back unexpectedly, away from a laptop entirely. What it also does is put the interview on a cloud service by construction: the recording, the transcript and whatever notes get generated all sit somewhere neither the reporter nor the source chose together, for as long as that account exists.

Reading the Mac's own audio instead

Coii AudioNotes records the reporter's microphone and the call's system audio as two separate tracks, without logging into the call as anything. Transcription runs segment by segment while the interview is still happening, and the summary is written on the same Mac once it ends.

The version of the interview a correction depends on

A pull quote in a published piece is a compression of forty minutes into one sentence, and the question a fact-checker actually asks is narrower than "what did they mean" — it's "what did they say, at that point in the call." A transcript sitting next to the recording answers that directly, at the timestamp it happened, rather than needing the reporter to reconstruct the exchange from a notebook written in shorthand mid-conversation.

Telling the reporter's voice from a source's

A bot inside a call separates speakers using the individual feed the platform hands it per participant. Reading the Mac's own microphone and system audio instead means that feed isn't available, so the separation works by voice print — matching each voice against the others in the same recording. A reporter and a single source are exactly two distinct voices, which is the case this works best against; a call with a press officer sitting in adds a third voice the same matching still handles.

Where this doesn't reach

It only records audio the Mac can actually hear. A video interview over a call running on the Mac is exactly that; a phone call taken entirely on a phone, with nothing routed through the Mac, is not, and neither is an in-person interview recorded on a separate handheld recorder rather than the laptop in the room. There's also no cloud archive across a beat's worth of interviews that a second reporter or an editor can search without being sent a specific file, no integration with a CMS, and it runs on macOS only — a newsroom split across Windows machines won't get one consistent tool from this alone.

Why the recording is the reporter's problem, not the source's

A source who's already careful about what they say on the record is one more thing more careful about who else might have access to the recording of it. A cloud notetaker's privacy page can describe its own retention and access policy, but that page is not something a source reads before an interview starts, and "who else can reach this" is a harder question to answer well when the honest answer involves a vendor neither party chose. Keeping the recording, the transcript and the summary on the machine that made the call sidesteps needing a good answer to that question at all.

An in-person interview, recorded with the laptop already in the room

Not every interview happens over a call. A reporter sitting across a table with a source, laptop open to take notes anyway, is already carrying the microphone the recording needs — there's no separate recorder to remember, charge or transfer files off afterward. The system-audio track stays effectively silent in that setup, which is expected: an in-person conversation has one track worth anything, the room the laptop's microphone can hear, and the transcript comes from that one track the same way it would from either side of a call.

What happens to an interview that never runs

Most interviews a reporter records don't end up as a published quote — background conversations, sources who end up not panning out, calls that get overtaken by the story changing shape. Those recordings still sit somewhere once the interview is over, and on a cloud notetaker that somewhere is an account on a server a reporter doesn't operate, for as long as the account exists, whether or not anything from the call is ever used. On a Mac, an interview nobody ends up citing sits exactly where every other interview does — a file on the reporter's own machine, deleted, kept or archived at the reporter's own discretion, never having existed anywhere else in the meantime.

A beat's worth of interviews, done one at a time

A single interview is where the fact-checking problem shows up first, but a beat with a dozen interviews on it surfaces a second one: which source, on which call, actually used a particular phrase, when two people described the same event slightly differently. A transcript sitting on the Mac, searchable by keyword, is the difference between a rough memory of who said what and being able to check the exact sentence before the piece goes out — without any of those interviews having passed through a shared index other stories also sit inside.

Checking one source's account against another's

A story built on several interviews eventually runs into the moment where two sources describe the same meeting or the same decision slightly differently, and the write-up has to reflect that rather than pick one version and hope. Working from memory of two separate conversations makes that comparison harder than it should be; two transcripts, each searchable on the reporter's own Mac, make it a matter of pulling up the relevant passage from each and reading them side by side, rather than trusting which account is remembered more clearly because it was the more recent call.

What an editor actually gets handed

An editor reading a draft rarely wants the full forty-minute recording — they want the two or three sentences the story's key claim rests on, findable without listening back through the whole call. A transcript makes that a search rather than a scrub through a timeline, which matters most on deadline, when the question isn't whether the interview happened but whether a specific sentence in it said what the draft says it said.

Who this genuinely isn't built for

A reporter working almost entirely by phone, calling sources from a handset rather than a laptop, won't get much from a tool that only hears what the Mac hears — that's a real limitation, not a small one, for that specific way of working. This is written for interviews that already happen on or through the Mac: video calls, in-person conversations recorded with the laptop in the room, and the write-up that follows either one.

The interview that runs long because it's finally going somewhere

An interview that's been guarded for the first twenty minutes sometimes opens up right when the scheduled time is supposed to end, and the worst moment to be doing math about a recording app's time limit or a notebook running out of pages is exactly then. Recording and transcription here run for as long as the conversation actually does, with nothing about the process changing whether the call ends on schedule or fifty minutes later than planned. The summary written afterward reflects the interview that happened, not a version trimmed to fit whatever length was expected going in — which matters most on precisely the calls where the useful part arrives late.

Notes taken during the call, kept separate from the transcript

Most reporters still take notes during an interview — flagging a moment to come back to, jotting the follow-up question a source's answer just raised. None of that has to change. The recording and its transcript exist alongside whatever's written in a notebook or a separate document during the call, not instead of it; the transcript is the backstop for the exact wording, and the notes stay what they've always been, the reporter's own read on which parts of the conversation actually mattered while it was still happening.

Setting it up is smaller than the decision to switch

The first launch asks for two macOS permissions — microphone access and permission to record system audio — the same prompt any Mac app requesting audio capture triggers. No account, no cloud workspace, nothing to connect before the first interview can be recorded. The larger adjustment is usually remembering to press record at all, rather than trusting a bot that used to join the call on its own.

What actually changes about the interview itself

Nothing about how the conversation runs changes — same questions, same platform, same source deciding how much to say. What changes is what exists once the call ends: a transcript searchable by keyword instead of a recording that has to be scrubbed through to find one sentence, and a short summary written the same day rather than reconstructed from memory once a draft is finally due. The interview is the same either way; what's available afterward, when a fact-checker or an editor comes asking, is not.

What it costs, what it runs on

Coii AudioNotes is $19, paid once, for three of a reporter's own Macs, with a 30-day trial needing no card and no account. It runs on macOS 13 Ventura or later, Apple Silicon or Intel. Against Otter's $8.33–$19.99-a-month tiers, a reporter doing a handful of interviews a week lands at $19 total rather than a subscription renewing for as long as the beat continues. The full comparison covers what Otter does that this doesn't, including the phone recording this page already conceded. For a technically inclined reporter already comfortable with a terminal, running an open-source Whisper script instead is free but leaves speaker separation, live transcription and the summary to be built by hand. A wider round-up of interview transcription tools for Mac and private meeting recorders covers the rest of the field, and the adjacent case for user researchers recording sessions and skipping the bot entirely covers nearby ground this page doesn't repeat.