There's a particular kind of dread that comes with finishing a good interview and realising you now have to turn an hour of recording into readable text. You could type it yourself, stopping and starting the audio every few seconds, losing the best part of an afternoon. Or you could upload it to one of the big transcription services and quietly hand over a recording of someone else's voice, someone who never agreed to that, to a company you've never met.
Neither option feels right, which is exactly why we built a third one. PeekoType now transcribes audio and video files entirely on your own PC, complete with automatic speaker labels, and it never sends a single second of the recording anywhere.
The short version: point PeekoType at a recording instead of a microphone, and it produces a full transcript with speakers labelled automatically, exportable as TXT, SRT, VTT, DOCX or PDF. Batch process a whole folder in one go. Everything runs locally. PeekoType is £39 once — file transcription is included, not a paid extra.
Point it at a recording instead of a microphone
PeekoType already does one thing brilliantly: it turns your live voice into text as you speak. File transcription is the natural next step, built on exactly the same on-device speech recognition. Instead of listening through your microphone in real time, it works through an existing file, whether that's a recorded interview, a meeting you captured, a lecture, or a podcast episode sitting in your downloads folder.
Drop in a file, or a whole folder of them, and PeekoType queues them up and transcribes each one in turn. There's no waiting around watching a single progress bar. Start a batch before lunch and come back to a folder full of finished transcripts.
Speaker detection: knowing who said what
A one-person voice memo is easy to transcribe. A thirty-minute conversation between three people is a different problem entirely, because a wall of text with no indication of who's speaking is barely more useful than the raw audio you started with. That's where automatic speaker detection comes in.
PeekoType identifies distinct voices in a recording and labels them automatically, so your transcript reads as a proper conversation rather than one long unbroken paragraph. It's built for exactly the situations where this matters most: interviews, meetings, panel discussions, focus groups, or a podcast with more than one host. If you regularly interview people for your work, this is very likely the single feature that saves you the most time.
Export it however you need it
A transcript is only as useful as the format it comes in, so PeekoType gives you five to choose from:
- TXT, for a clean, plain-text transcript you can paste anywhere.
- SRT and VTT, the standard subtitle formats, ready to drop straight into a video editor.
- DOCX, for a properly formatted Word document you can edit, mark up or send on.
- PDF, print-ready and shareable exactly as it appears.
Whether the transcript is heading into a written article, a subtitle track, a court bundle or a research paper, there's a format here that fits without any manual reformatting.
Why offline transcription matters more than people realise
Cloud transcription services are everywhere, and most of them are perfectly competent. What they all have in common is that your recording, and everyone's voice on it, has to leave your computer to be processed. For a casual voice memo, that might not trouble you. For a client interview, a confidential meeting, a legal deposition, or a patient consultation, it's a genuine problem, and often one that other people on the recording never had the chance to consent to.
Our guide to GDPR-compliant voice typing covers this in more depth, but the short version is that on-device processing sidesteps the whole issue. There's no server storing your recording, no third party processing it, and nothing to disclose in a data processing agreement, because nothing ever leaves your machine. That's precisely why professionals in legal and clinical work tend to steer well clear of cloud-based transcription tools altogether.
Who this is built for
| Who you are | What offline transcription solves |
|---|---|
| Journalists and interviewers | Turn recorded interviews into text with speakers labelled, without sending sources' voices anywhere |
| Meeting-heavy teams | Batch-transcribe recorded meetings into a searchable, shareable record |
| Students and researchers | Transcribe lecture recordings or interview data for a dissertation, kept private |
| Content creators | Generate SRT subtitles for video, or a clean transcript for a podcast show notes page |
| International teams | Transcribe multilingual meetings without a third party ever hearing the recording |
How it fits alongside everything else
File transcription is part of PeekoType rather than a separate product. One £39 payment covers all of it: dictation, custom vocabulary and text snippets, translation and read-aloud voices, plus batch transcription, speaker detection, and the five export formats above. There is a free 14-day trial with no card required and nothing held back, so you can transcribe a real recording of your own before deciding.
It also pairs neatly with our new live captions for Teams and Zoom, if you want the conversation captured in real time as well as transcribed properly afterwards.
The bottom line
Transcription shouldn't mean choosing between hours of manual typing and handing a recording of someone's voice to a stranger's server. PeekoType gives you a third option: fast, accurate, labelled by speaker, and entirely private, because it never leaves your PC in the first place.
Start your free trial from our homepage, or email support@peekotype.com if you'd like to talk through whether it suits the kind of recordings you work with.