You have a dictaphone recording — a dictated brief, an interview, clinical notes — and you need it as text. Here's how to do it without uploading a single second of audio: export the file from your recorder, drop it into Whisper Notes, and Parakeet V3 transcribes it entirely on your iPhone or Mac. No cloud, no account, no per-minute fees. $6.99 once.
Who Still Uses Dictaphones — and Why It Matters
The dictaphone never went away. It just moved into professions where speaking is faster than typing and the content is too sensitive to lose — or leak.
- •Lawyers dictate briefs, attendance notes, and letters between meetings. A solicitor's dictation may contain client names, case strategy, and privileged detail that must never touch a third-party server.
- •Doctors dictate clinical notes and referral letters after each patient. That audio is medical data in its rawest form.
- •Journalists record interviews on voice recorders and phones. Protecting a source starts with not uploading their voice to someone else's cloud.
- •Researchers collect hours of field interviews and observations, often under ethics agreements that explicitly restrict where recordings can be processed.
For all four groups, the bottleneck is the same: turning hours of dictation into text. Traditionally that meant a typist, an outsourced transcription bureau, or a cloud service — each one a person or a server that hears everything you said. Offline transcription removes the middleman entirely.
From Recording to Text in Three Steps
1. Export the file from your dictaphone
Connect your voice recorder over USB (or use its companion app) and copy the recording to your Mac or iPhone. Most digital dictaphones — Olympus, Philips, Sony, Zoom, or the Voice Memos app on your phone — save recordings as MP3, WAV, or M4A. Any of these work as-is; no conversion needed.
2. Import into Whisper Notes
Drag the file into Whisper Notes on Mac, or share it to the app on iPhone. There's no length limit — a two-minute memo and a three-hour interview are both fine. Video files work too: import an MP4 or MOV and the app transcribes the audio track.
| Format | Type | Supported |
|---|---|---|
| MP3 | Audio — most digital dictaphones | Yes, any length |
| WAV | Audio — uncompressed recorders | Yes, any length |
| M4A | Audio — iPhone Voice Memos | Yes, any length |
| MP4 | Video — audio track transcribed | Yes, any length |
| MOV | Video — audio track transcribed | Yes, any length |
3. Transcribe locally, export anywhere
Hit transcribe. Parakeet V3 — the default model — processes the audio on your device's own chip, roughly 10× faster than Whisper, with a 6.32% word error rate on clear speech. The result is a transcript in timestamped paragraphs: click any paragraph to jump to that moment in the recording, which makes verifying a quote or a dosage instruction take seconds instead of minutes.
An imported recording, transcribed with timestamps — click any segment to hear the original audio
When you're done, export as TXT for a document, or SRT/VTT with timestamps if the recording came from a video. Edit inline first if you like — corrections stay synced to the audio.
Why Offline Beats Cloud for Dictation
Cloud transcription services work by uploading your audio to their servers, processing it there, and storing it under their retention policies. For a podcast episode, fine. For a dictated client brief or a patient note, that's a confidentiality problem you don't need to have.
With on-device transcription there is nothing to secure, because nothing leaves. No upload, no account, no server logs, no data processing agreement to negotiate. For European professionals, this is GDPR-friendly by architecture: you're not transferring personal data to a processor, because there is no processor.
| Offline app (Whisper Notes) | Cloud transcription services | |
|---|---|---|
| Where audio is processed | On your own device | Vendor's servers |
| Confidentiality | Audio never leaves the device | Depends on vendor's policies |
| GDPR | Friendly by architecture — no data transfer | Requires processing agreements |
| Cost | $6.99 once | Monthly subscription or per-minute fees |
| Works without internet | Yes — courtroom, clinic, plane, field site | No |
| Account required | No | Yes |
The economics matter for heavy dictation users too. If you dictate an hour a day, per-minute cloud pricing adds up fast, and subscriptions bill you whether you dictate or not. A one-time purchase costs the same whether you transcribe ten minutes a month or ten hours a week.
Getting the Most Accurate Transcript
Two settings make the biggest difference for professional dictation:
Custom vocabulary. Speech models are trained on general language, so they stumble on exactly the words that matter most in your field — client surnames, drug names, statute citations, technical jargon. Whisper Notes lets you add these terms to a custom vocabulary, and the model uses them to resolve ambiguous audio. Add your ten most-dictated proper nouns and you'll eliminate most recurring errors.
Model choice by language. Pick the model that matches your dictation language:
| English / European | Parakeet V3 — 25 European languages, 6.32% WER, 10× faster than Whisper, only 465MB |
| Chinese / Japanese / Korean | SenseVoice — fastest for CJK and Cantonese, 52× real-time |
| Other languages | Whisper Large V3 Turbo — 100+ languages, ~1.5GB, slower but broadest coverage |
And one habit from the analog dictation era still pays off: hold the recorder close, speak at a steady pace, and dictate punctuation-friendly sentences. Clean audio in, clean text out.
Frequently Asked Questions
Can I transcribe old dictaphone recordings?
Yes. Copy the files off your recorder and import them into Whisper Notes — MP3, WAV, and M4A all work, at any length. A recording from ten years ago transcribes exactly like one from this morning; accuracy depends on the audio quality, not the age of the file.
What is the best app to transcribe dictaphone recordings?
Judge any app on four criteria: where the audio is processed, which file formats it accepts, whether it handles domain jargon, and what it costs over a year of use. If your dictation is confidential — legal, medical, journalistic — we'd argue on-device processing is non-negotiable, and that's exactly why we built Whisper Notes: local transcription, MP3/WAV/M4A/MP4/MOV import at any length, custom vocabulary, $6.99 once.
Does dictaphone transcription work without internet?
With Whisper Notes, yes — fully offline. The speech models are downloaded once and run on your iPhone or Mac's own chip, so you can transcribe in a courtroom basement, on a plane, or at a field site with no signal. No connection is ever required for transcription.
How accurate is offline dictaphone transcription?
Parakeet V3 achieves a 6.32% word error rate on clear audio — competitive with cloud services. The remaining errors cluster around proper nouns and specialist jargon, which is what the custom vocabulary feature is for: add your recurring names and terms, and accuracy on exactly those words improves markedly.
Download for iOS