An interview conducted for a research project isn't just a conversation to be summarised — it's data that an ethics board has usually already asked about, in the application that got the study approved before the first interview was scheduled. Where that recording ends up afterward, and who else can reach it, is a question that has to be answered honestly on the consent form a participant signs, which makes the recording tool part of the research design rather than an afterthought bought after the fact.
What a cloud transcription service is honest about
Otter's own terms describe a cloud service: the recording is uploaded to produce a transcript, and the account holding it exists on infrastructure Otter operates, not the researcher. That's not a hidden fact — it's stated plainly — and for a lot of research it's a fine trade for a transcript produced quickly and a mobile app that can record an interview that starts unexpectedly, away from a laptop. What it means for a study with a strict consent form is that "where does the recording go" has an answer that names a vendor, which some ethics review processes are more comfortable with than others.
Recruiting participants who ask where the recording goes
A recruitment email or a study flyer sometimes gets a question back before a participant agrees to be interviewed at all — where exactly the recording will be stored, and who besides the researcher can access it. "On my own laptop, and nowhere else" is a complete, honest answer to that question, and one a participant doesn't need to trust a third-party vendor's privacy policy to evaluate. It's a small thing in a recruitment email, but it's occasionally the detail that decides whether a hesitant participant agrees to the interview at all.
An interview transcribed without leaving the Mac
Coii AudioNotes records the researcher's microphone and, for a call, the system audio, as two separate tracks, and transcribes segment by segment while the interview is happening. None of it — the recording, the transcript, the summary — is uploaded to produce any of the three.
What a transcript settles that field notes alone can't
A researcher taking notes during an interview is already interpreting as they write — deciding in the moment which phrase is worth capturing verbatim and which can be paraphrased. A full transcript removes that choice from the moment of the interview itself: the exact wording is there afterward, to quote precisely in a paper or to re-read against a second interview, without depending on which sentence the researcher's hand happened to keep up with.
Where this genuinely stops
There's no integration with a qualitative coding tool — no direct export into NVivo, Atlas.ti or a similar package, so a transcript produced here has to be brought into that software the same way a transcript from anywhere else would be. There's no shared workspace a co-author or a research assistant can browse without being sent the file directly, and it runs on macOS only, so a lab with a mix of Mac and Windows machines won't get one consistent tool from this alone.
Revising the interview guide between sessions
A study conducted over several weeks or months often has the researcher revising the interview guide slightly as patterns emerge — a question that keeps landing awkwardly, a follow-up worth adding after an earlier participant raised something unexpected. That revision works better with a transcript to look back on than with memory of "something felt off about question four in a couple of the early interviews," because the transcript shows exactly where and how it landed awkwardly, not just that it did.
What the recording's home says on a consent form
"The recording is transcribed on my own laptop and not uploaded to any service" is a sentence a participant can actually evaluate, in a way that "the recording is processed by our transcription vendor, subject to their privacy policy" usually isn't for someone reading a consent form for the first time. Neither sentence is a certification of anything — that's a matter for the researcher's own ethics board — but the first one is shorter, and it's true regardless of which vendor's terms happen to be in effect the week the study runs.
A transcript for a study that runs for months, not one interview
A single interview transcribed the same day it happened is useful immediately; a study built on twenty or thirty of them over a semester surfaces a different need — finding which participant, across which interview, used a specific phrase, when two accounts of the same event differ. Reconstructing that from field notes across a full study isn't realistic past the first several interviews; transcripts sitting on the researcher's own Mac, searchable by keyword, make it a matter of pulling up the right passage rather than trusting which interview is remembered most clearly.
The interview that runs long because the subject just opened up
An interview guarded for the first half hour sometimes turns candid right as the scheduled time runs out, and stopping to manage a recording app's time limit at exactly that moment is a bad trade against the actual data. Recording and transcription here run for as long as the conversation does, so the extra twenty minutes that turned out to matter most get captured the same way as the rest of it.
Comparing what two participants said, without trusting memory
A study built on interviews eventually runs into the moment where two participants describe the same event, or answer the same question, noticeably differently — and the write-up has to reflect that honestly rather than defaulting to whichever account is remembered more clearly. Two transcripts, both searchable on the researcher's own Mac, turn that comparison into pulling up the relevant passage from each and reading them side by side, instead of relying on which interview happened more recently and is therefore easier to recall.
A pilot interview, before the study is finalised
Most studies run at least one pilot interview to test whether the questions actually work before the full round begins, and a pilot is exactly the kind of recording a researcher wants to review closely — the places a question landed awkwardly, the follow-up that should have been asked but wasn't. A transcript of the pilot, produced the same day it was recorded, makes that review a matter of reading rather than relistening, while there's still time to revise the interview guide before the participant pool is used up.
What happens to a recording once the study is over
An interview recorded for a study doesn't stop being data once the paper is published — many ethics protocols specify a retention period, and some studies get revisited years later when a new question arises about the original data. A recording that never left the researcher's own Mac in the first place is exactly where it's always been when that question comes up, rather than a file the researcher has to request back from a vendor whose retention practices may have changed since the study ran.
Writing up a finding without misquoting a participant
A paper or a report that quotes a participant directly carries a real cost if the quote is slightly wrong — a paraphrase remembered as exact wording, or a phrase attributed to the wrong participant across a study with a dozen similar conversations. A transcript checked against the recording before a quote goes into a draft is a small extra step that avoids that specific, avoidable error, and it's a step that only works if the transcript exists to check against in the first place.
Setting it up is smaller than the ethics application
The first launch asks for microphone access and permission to record system audio — the standard prompt any Mac app requesting audio capture triggers. No account, no institutional sign-on, nothing to configure before the first interview can be recorded. The heavier lift is usually the ethics application itself, which this doesn't shorten, only makes one sentence of it easier to write honestly.
A second interview, months after the first
Longitudinal research sometimes brings a researcher back to the same participant months or years after a first interview, and the value of that second conversation depends partly on being able to check it against the first one — did this person's account of the same event change, and if so how. That comparison only works if the first interview's transcript is still sitting somewhere findable when the second one happens, which is exactly what a searchable archive on the researcher's own Mac provides, without depending on whichever vendor hosted the first transcript still existing, in the same form, years later.
Transcribing a group interview or a focus group
Not every qualitative interview is one-on-one. A focus group with four or five participants talking, sometimes over each other, is a harder case for voice-print separation than a single subject is — more voices in the same recording means more room for two similar-sounding ones to be confused, and a transcript from a group session usually needs a closer read-through to confirm who said what before it's relied on for a direct quote. It's still transcribed on the same Mac, without being uploaded to do it; it just deserves more scrutiny before a specific speaker attribution is trusted.
Who this genuinely isn't built for
A large, multi-site study with a research assistant transcribing interviews a principal investigator never personally conducts, or a lab that needs transcripts landing directly in a shared coding tool without anyone moving a file by hand, has a real workflow this doesn't replace. This is written for the researcher — often working alone or with one or two collaborators — who conducts and later analyses their own interviews, and who would rather answer "where does the recording go" with a sentence about their own laptop than a link to a vendor's privacy policy.
What it costs, what it runs on
Coii AudioNotes is $19, paid once, for three of a researcher's own Macs, with a 30-day trial needing no card and no account — long enough to run it through a full round of pilot interviews before a study starts in earnest. It runs on macOS 13 Ventura or later, Apple Silicon or Intel. Against Otter's $8.33–$19.99-a-month tiers, checked today, a study that runs for a semester or a year lands at $19 total rather than a subscription renewing for as long as the fieldwork does. The full comparison covers what Otter does that this doesn't, including the mobile recording this page already conceded. For a researcher already comfortable running a script, comparing this against an open-source Whisper setup and against MacWhisper covers the local-first alternatives directly, and the adjacent cases for user researchers and journalists cover nearby interview-based work this page doesn't repeat. A wider round-up of interview transcription tools for Mac covers the rest of the field, and the case for skipping a meeting bot entirely applies here too, for the interviews that happen over a call.