A meeting has an agenda. An interview has a source.

The tools built for meetings mostly work for interviews too, which is why so many people reach for whatever notetaker they already have. The differences show up later, and they are specific: a meeting summary that paraphrases a decision is fine, but an interview transcript that paraphrases a quote is a different document than what was actually said. A meeting recording sitting on a vendor's server is a minor question for most teams; an interview recording sitting there, when the person on the other end agreed to speak on the understanding it stayed between the two of you, is not minor at all. Six tools below, plus how each one answers those two questions specifically.

There is a third difference worth naming before the list: an interview is usually two people, not a room, and the thing worth capturing precisely is not a decision the group reached together but a specific sentence one person said in a specific way. A meeting summariser that smooths a rambling answer into a tidy bullet point is doing its job for a meeting and actively working against a journalist or a researcher who needs to know exactly what was said, hedges and false starts included, because a paraphrase of a quote is not a quote. That single distinction is why "good at meetings" and "good at interviews" are not the same recommendation, even when the underlying transcription engine is identical.

There is also a practical difference in how the recording itself gets made. A meeting usually has a fixed structure to work with — a calendar invite, a video call already open, a bot that can be scheduled ahead of time. An interview is more often improvised: a phone call taken standing up, a conversation in a coffee shop, a video call joined from a source's personal account rather than a company one a bot has permission to enter. Tools built around a calendar and a scheduled bot handle the first situation well and the second one poorly, which is part of why several of the names below lean toward capturing whatever the device can hear rather than assuming there is always a meeting link to attach to.

Otter — fast turnaround, with a bot standing in for a recorder

Otter's Pro and Business tiers run $8.33 to $19.99 a month annually, per its pricing page checked today, and its strength for interviews is speed: a live transcript appears while the conversation is still happening, useful for glancing at a name or a figure mid-conversation rather than waiting until afterward. For a scheduled video call, its bot joins Zoom, Teams or Google Meet directly rather than capturing audio from the Mac itself. Its privacy page states it de-identifies audio before using it to train its own models — meaning training happens, on de-identified data, which is worth knowing before recording a source who was told the conversation was private. Its free Basic tier caps out at 300 transcription minutes a month, which a longer interview or a series of them can use up faster than a typical business meeting would. Full comparison here.

Descript — for turning the interview into the thing you publish

Descript's pricing page, checked today, runs Free at $0 through Hobbyist at $16–$24, Creator at $24–$35 and Business at $50–$65 a month, with a desktop app requiring macOS 14 or newer. Nothing else on this page is built to cut an interview transcript directly against its audio and video for a podcast episode or an article's embedded clip — that timeline-based editing is Descript's actual product, with transcription as the way in rather than the end point. It is entirely cloud-based, with storage limits scaling from 5GB free to 2TB on Business, so the recording lives on Descript's infrastructure for as long as the project does. Its "Rooms" feature can host up to ten participants in a remote recording session, wider than a typical two-person interview needs, but useful for a panel discussion or a group interview recorded the same way. Full comparison here.

MacWhisper — batch through a backlog of old interviews, locally

MacWhisper Pro is €64, paid once, confirmed on its own site today, and its batch-processing feature is the reason it belongs on this list specifically: a folder of forty old interview recordings can be transcribed in one pass, locally, with automatic speaker labelling and export to SRT, VTT or plain text. It also handles YouTube and podcast audio directly, useful for transcribing somebody else's published interview rather than one you recorded yourself. Summaries route through a connected AI provider the user picks, which is a setup step worth knowing about before assuming the whole pipeline is local by default. Its export formats — SRT and VTT alongside plain text — matter specifically for anyone producing a video interview that needs captions, a job neither the meeting-first tools nor Coii AudioNotes are built to do. Full comparison here.

superwhisper — quick capture for an interview taken on the move

superwhisper's free tier already includes meeting recording and transcription, with Pro at $8.49 a month adding more on top, per its pricing card checked today. Its offline mode, strongest on Apple Silicon Macs, matters for a reporter recording in a location without reliable signal — a real scenario for fieldwork or an in-person interview away from a desk. It is built primarily for dictation first, with meeting and interview capture as a second feature, which shows if the workflow leans more on transcription depth than on typing text into other apps. For fieldwork specifically — an interview conducted somewhere without a reliable connection — its offline mode is the only one on this page, besides the two fully local tools, that was built with that scenario in mind rather than as a fallback. Full comparison here.

Notta — a bot or no bot, by choice, for a scheduled call

Notta's pricing page, checked today, lists Free at $0 with 120 minutes a month, Pro at $8.17 a month annually, and Business at $16.67, with recording available either through a bot that joins the call or a bot-free mode that records on the device instead — a choice most competitors on this page do not offer explicitly. Its own privacy documentation states that, depending on the plan, third-party speech-recognition partners may use audio data for training, with no settings page describing an opt-out — worth weighing against a source who was promised the recording stays between the two of you. Having both a bot-based and a device-based recorder in the same product is useful for someone who interviews both over video calls and in person and would rather not run two different apps for the two situations. Full comparison here.

Fireflies — a written no-training line, useful for protecting a source

Fireflies' plans, detailed on its pricing page checked today, run from free through Pro and Business tiers, and its site states plainly that customer data is never used for AI training or for any purpose outside direct business needs, with a stated zero-data-retention policy alongside it. For an interview where the source's identity matters, that written commitment is a real answer, even though the recording still reaches Fireflies' servers before the policy applies to it. A bot joins the call by default; there is no bot-free mode. Its search across a growing archive of past interviews is a real advantage for anyone building a beat over months or years, where finding the one thing a source said eight interviews ago matters more than any single transcript on its own. Full comparison here.

Coii AudioNotes — the recording that never has a server to reach

Coii AudioNotes is $19, paid once, for three of your own Macs, with a 30-day trial and no account required. For a two-person interview it captures the microphone and system audio as separate tracks, transcribes while the conversation runs, and tells the two speakers apart by voice print. The summary — decisions, follow-ups, the moment worth going back to — is written by a language model bundled into the app, so nothing about finishing the interview requires a network connection at all.

It has no editing timeline like Descript, no bot for a call nobody remembered to record, and it runs on nothing but a Mac. Full comparison here for the closest fast-turnaround alternative.

Verbatim is a setting, not a given

Meeting-first transcription tools are generally tuned for readability — filler words trimmed, false starts smoothed, punctuation inferred rather than dictated — because that is the right default for a meeting, where nobody wants to reread every "um." It is the wrong default for a quote that is going to be published, where the exact wording is the thing being checked. Before treating any transcript on this page as publication-ready, read it against the audio at the specific moment a quote comes from, rather than trusting a cleaned-up transcript's phrasing at face value — a habit worth building regardless of which of the six tools produced it, or how accurate its transcript generally is. This is not a criticism of any single product on this page; it is a mismatch between what most transcription tools optimise for by default and what a published quote actually requires, and the fix is a five-minute check rather than a different piece of software.

What actually decides it

If the recording is going to become a published piece — audio, video, or a transcript with cuts — Descript is built for that step in a way nothing else here attempts. If speed while the conversation is still running matters more than anything else, Otter's live transcript is hard to beat. If the honest answer is that a source was promised privacy and that promise should not depend on reading a vendor's opt-out policy correctly, MacWhisper and Coii AudioNotes are the two names on this page where the recording never leaves the Mac to begin with.

Price is a smaller factor here than in most comparisons on this site, since even the priciest name — Descript's Business tier at $65 a month — is a rounding error against what a freelance journalist or researcher bills for a single piece built on the interviews it transcribes. The sharper trade-off is speed against control: Otter's live transcript saves time during the conversation itself, while MacWhisper's batch mode and Coii AudioNotes's finished minutes save time afterward, when the actual writing or the actual write-up starts. Neither is wrong; they are simply optimised for different points in the same process, and it is worth being honest about which point actually causes the most friction in a real workflow before picking based on price or platform alone.

The table

Which to actually try

Run the same short interview through two candidates before committing — the difference that matters most, accuracy on names and technical terms, only shows up on real audio, not a vendor's demo clip. If the piece is going to be published, start with Descript; if the promise made to the source is the thing that has to hold, try Coii AudioNotes for thirty days and see whether a local transcript with no bot and nothing uploaded actually changes how the conversation feels on the other end. That last point is easy to underrate until it is tested directly: a source who is told, honestly, that the recording never leaves the Mac in front of them often says more than one who was only told the app "takes privacy seriously," because the first is a fact and the second is a slogan every vendor on this page could print.