Dictation for journalists
Short answer: For journalists, Parakeety is on-device push-to-talk dictation for the writing side of the job: drafting the story, typing up your interview notes by voice, getting a quote into the document while the phrasing is fresh. The audio is transcribed on the Mac and discarded, so unpublished material and source-identifying notes never touch a third-party server, and there is no cloud account holding your work for anyone to compel. Reporters have had two unhappy options on Macs. Cloud dictation gets words off the laptop fast but routes everything you say through someone else’s servers. Apple’s built-in dictation cuts off mid-sentence and was never built for sustained note-taking. Parakeety is a third path: local push-to-talk dictation for Mac, part of the wider move toward dictation where the audio never leaves the laptop. Hold the section key, talk through the paragraph or the note, release; the words paste at the cursor in whatever you were writing in.
Where dictation fits in a reporter’s day
The honest framing is that dictation earns its keep on first drafts and on note-writing, not on editing. Once you are moving sentences around and tightening a piece, that belongs to the keyboard and the cursor. Where voice wins is the moment the words are still forming: the framing paragraphs of a feature, a quick reaction piece against a deadline, the running notes you write up the second you leave a doorstep or hang up the phone.
That maps to specific points in the day:
- Writing up an interview. Talking through what was said, in your own words, with the exact quotes you remember, straight into the working document while it is still sharp in your head.
- Drafting against a deadline. Getting the structure and the framing paragraphs of a longer piece down before they evaporate, then editing on the keyboard.
- Field notes. Dictating observations into a note on the laptop the moment you are back at it, rather than typing into a phone or trusting memory until you get to a desk.
- Quotes and attributions. Reading a quote aloud from your shorthand into the document so the wording lands exactly as you took it down.
The general first-draft case across writing work is covered in local dictation for writers and researchers; this page is the reporting-specific version, where the privacy of the source notes is doing as much work as the speed.
The architectural privacy story
Source protection is the load-bearing duty in reporting. The material that matters most, the name of a source, the notes that would identify them, the draft of a story before it runs, is exactly the material you least want sitting on a server you do not control. Cloud dictation services move the audio off your machine to transcribe it, and usually retain the transcript for some period to run the product. Whether that is consistent with your obligation to a source, your outlet’s policy, or an embargo, is your judgment to make.
Cloud services answer this with a contract: a data-processing agreement, a retention promise, an audit log. That suits some newsrooms. The cleaner answer to the same problem is to not transmit the audio at all. Parakeety processes audio on the Apple Neural Engine, in memory, in around 0.2 seconds for a ten-minute dictation, then discards the buffer. There is no third-party processor in the data-flow diagram because there is no network call at transcription time. Nothing is written to disk and nothing is logged.
That difference matters most when someone comes looking. There is no Parakeety account to sign into and no cloud store of what you dictate, so there is no third-party-held copy of your drafts or notes for a subpoena to reach. The distinction between this architectural answer and the contractual one, where audio leaves the device but the vendor promises to handle it well, is unpacked in the cornerstone piece on architectural versus contractual privacy. It was written for the healthcare audience, but the legal-privilege section reads across cleanly to reporting, and local dictation for lawyers and solicitors restates the same argument where privilege rather than source protection does the heavy lifting.
Cloud dictation versus local, for source-sensitive work
A scan-first comparison of where the unpublished material and the source notes end up:
| Question | Cloud dictation | Parakeety (local) |
|---|---|---|
| Where audio is transcribed | On the vendor’s servers | On the Mac, on the Apple Neural Engine |
| Account required | Usually yes, with a login | No account, no login |
| Server copy of your text | Often retained for a period | None; nothing is uploaded |
| What a subpoena to the vendor reaches | Whatever the vendor holds | License records only, never drafts or notes |
| Works offline on a flight or in the field | No, needs a connection | Yes, after the first-launch model download |
Practical fit
- It pastes wherever the cursor is: Google Docs, Word, Pages, Scrivener, Notes, Slack, email, and any browser-based content-management system or newsroom editor. The integration is at the operating-system level, so there is no plugin to install in your CMS.
- Each license covers one Mac. A reporter’s own laptop, or a shared machine on the desk, both shapes are fine; the model is one license per machine, with no per-user or seat-based subscription.
- The model auto-detects the language across 25 European languages, which matters for foreign-desk work and for getting a quoted line from a non-English source down verbatim.
- The push-to-talk key is currently the section key (§, below Esc). A user-configurable shortcut is on the roadmap.
- First launch downloads the speech model (around 600 MB) and asks for microphone and accessibility access. After that the app lives in the menu bar and runs offline, which is the on-a-flight, on-a-train, in-a-field case.
What Parakeety is not
Worth being plain about, because the reporting use case invites the misunderstanding. Parakeety does not transcribe recorded interviews. It is push-to-talk: you hold a key, talk live, and release. It does not accept an audio file or a recording as input, so it will not take the audio of an interview you recorded on a phone or a recorder and hand you back a transcript. That is a different product category. For transcribing a recorded interview, MacWhisper is the conventional Mac tool, and it is built for exactly that file-in, text-out job.
Parakeety is for the writing, not the recording. Use it to draft the story, to write up your notes by voice, to get a remembered quote into the document. It also does not do AI cleanup or rewriting; the transcript that pastes is what the model produced, with the punctuation it inferred. That tends to be good at the sentence level and lighter on paragraph breaks across longer passages, which you add in the edit pass. For the first-draft and note-writing job it is built for, that trade is fair.
FAQ
- Can Parakeety transcribe a recorded interview?
- No. Parakeety is push-to-talk dictation, not file transcription. You hold a key, talk live, and release; the words paste at the cursor. It does not accept an audio file or a recording as input, so it will not turn a recorded interview into a transcript. If your need is to transcribe recorded interview audio, MacWhisper is the conventional Mac tool for that workflow. Parakeety is for the writing: drafting the story, typing up your notes by voice, getting a quote into the document while you remember the exact phrasing.
- Does any unpublished material leave my Mac?
- No. Audio is captured to a memory buffer while you hold the key, transcribed on the Apple Neural Engine, pasted at the cursor, and the buffer is discarded. Nothing is written to disk by Parakeety and nothing is logged. The only outbound calls the app makes are the one-time speech-model download on first launch and periodic license checks, which carry the license key, a hardware ID hash and the machine hostname, never audio or text. There is no cloud account holding your drafts and no server-side copy of source-identifying notes.
- Is there an account or login that could be subpoenaed?
- There is no Parakeety account to sign into and no cloud store of what you dictate. Because the transcription happens on the device and the audio is discarded, there is no third-party-held record of your dictation for anyone to compel. A subpoena directed at Parakeety would find license records, not story drafts or notes, because drafts and notes are never sent to us. Your own files on your own Mac are still subject to whatever legal process applies to you directly; Parakeety simply does not add a third party that holds a copy.
- Does it paste into Google Docs, Word and a CMS?
- Yes. Parakeety pastes the transcript by synthesizing a paste keystroke into whichever app has focus, so it works in Google Docs, Word, Pages, Scrivener, the system Notes app, Slack, email, and any browser-based content-management system or newsroom editor. The integration is at the operating-system level rather than the app level, so there is no plugin to install in your CMS and no permissions to grant inside it.
- Can I dictate foreign-language quotes from a source?
- Yes, across 25 European languages with automatic detection. You can draft an English paragraph, switch to dictating a French or German quote verbatim, then switch back to English, without changing any setting. The model identifies the language by itself. Languages outside the European list, including Mandarin, Japanese, Korean, Arabic, Hindi, Turkish and Hebrew, are not supported at the moment.
Try it
Parakeety is a Mac menu-bar app. Hold the section key, talk, release; your words paste at the cursor in whichever app you were typing into. Audio never leaves the machine, so there is no server copy of your drafts or source notes to worry about. There is a free 7-day trial with no card required. After that it is $30 once.