Recording a thought takes three seconds. Finding it again in six weeks is the hard part.

That’s the whole reason transcription matters. An untranscribed voice note is a forty-second audio file with a timestamp for a name, and you will never listen to it again. A transcribed one is searchable text.

So: which app actually turns speech into text you can use?

How this list is sourced

I have not personally run every app here for a month, and you should distrust any blog that claims it has.

What I have done is verify the checkable parts. Every app name, developer, star rating and rating count below was pulled from the US App Store on 11 August 2026. Feature descriptions come from each app’s own store listing and developer documentation.

I’ve kept prices out of the table on purpose. Most of these are free to download and charge through subscriptions that change by region and by promotion. The two genuine up-front purchases are flagged in the text, because that’s a real structural difference, not a number that rots.

First, clear up the Whisper confusion

This trips up almost every comparison article on this topic, including the previous version of this one, so it’s worth doing properly.

Whisper is a speech recognition model made by OpenAI. It’s released under the MIT licence on GitHub. It is genuinely open source, it is genuinely free, and it is genuinely multilingual. It is not an app. You cannot install it from the App Store. Running it yourself means a command line, Python and a machine with enough headroom.

Whisper Transcription is a different thing entirely. It’s a commercial app by Jordi Bruin, published under Good Snooze, and as of 11 August 2026 it sits at 3.8 stars across 156 ratings on the US App Store. It’s free to download, with paid “Pro” and “Assistant” in-app purchases. It runs on iPhone, iPad and Mac, and it can download a model to transcribe locally on-device.

Both are fine. They are not the same product, and articles that list “Whisper — free, open source” in a table of consumer apps are describing something you cannot actually download.

There are several other apps built on the same underlying approach. Aiko by Sindre Sorhus is a $24.00 up-front purchase at 4.7 stars across 325 ratings. Superwhisper - AI Dictation by SuperUltra, Inc. is 4.4 stars across 808 ratings.

The honest summary: the model is excellent, the apps wrapping it vary enormously, and the rating counts here are small enough that you should treat every one of them as less battle-tested than Otter or Apple’s own tools.

Second, Apple’s transcription is in two places

Another thing comparison tables get wrong. People say “Apple Notes transcribes now”, and the answer is: yes, but so does Voice Memos, and they work differently.

Voice Memos transcribes recordings and lets you search across them. Apple documents viewing and copying a Voice Memos transcript, including automatic transcription of older recordings made before the feature existed. Searching the recordings list matches text in the audio, not just the title. Apple notes the feature isn’t available in every country or region.

Notes does something different: it records audio inside a note and transcribes it alongside your typed text. Apple’s guide to recording and transcribing audio in Notes states it requires iPhone 12 or later and one of a specific list of device languages, and that you can tap any line of transcript to jump the audio to that point.

If you want a plain voice recorder with searchable transcripts, that’s Voice Memos. If you want the recording attached to written context, that’s Notes. Both are free, already installed, and process on-device.

Voice Memos sits at 4.8 stars across 1,060,385 ratings, which is more ratings than every other app on this page combined. Worth trying before you install anything.

Pick your category first

There are really only four situations, and the right answer barely overlaps between them.

  • Meetings and interviews with several people → you need speaker labels, which almost nothing here has
  • Notes to yourself while walking or driving → you need speed of capture, not accuracy
  • Anything confidential → you need on-device processing, which rules out most of the good stuff
  • Voice as an input method rather than a record → you want dictation, not a recorder

Quick comparison

Developers, ratings and rating counts verified on the US App Store, 11 August 2026.

AppDeveloperRatingRatingsProcessingSpeaker labels
Voice MemosApple Inc.4.81,060,385On-deviceNo
NotesApple Inc.4.8624,281On-deviceNo
OtterOtter.ai, Inc.4.871,711CloudYes
Wispr FlowWispr AI INC4.813,308CloudNo
VoicenotesCoping Hard Inc.4.86,589CloudNo
RevRev.com, Inc.4.54,204CloudYes
Just Press RecordOpen Planet Software Limited4.31,520On-deviceNo
NottaMind Cruiser Limited4.11,384CloudYes
SuperwhisperSuperUltra, Inc.4.4808On-device optionNo
AikoSindre Sorhus4.7325On-deviceNo
Whisper TranscriptionJordi Bruin3.8156On-device optionNo

Read the rating counts, not just the stars. A 4.7 across 325 ratings and a 4.8 across a million are not the same claim.

If you record meetings

Otter — 4.8 stars, 71,711 ratings

The one everybody names, and the reason is speaker identification. It labels who said what, transcribes live while the meeting is happening, and connects to the usual video conferencing tools.

There is a free tier and a paid tier. Both the free minute allowance and the paid price have changed more than once, and I’m not going to print a figure that’s wrong by the time you read this — check otter.ai’s pricing page.

The real trade-off: it uploads your audio. That is not a scandal, it’s how the product works, but it means Otter is the wrong tool for a conversation you promised to keep private.

Rev — 4.5 stars, 4,204 ratings

From Rev.com, who have been selling human transcription for years. The app does automated transcription, and the same account can escalate a file to a paid human transcriptionist when the automated pass isn’t good enough.

Worth knowing about if you occasionally need transcription that is actually accurate rather than mostly accurate, for a legal or published use.

Notta — 4.1 stars, 1,384 ratings

Multi-speaker transcription with a lower rating and a much smaller rating count than Otter. Listed here for completeness rather than as a recommendation.

If you’re capturing notes to yourself

Voice Memos — 4.8 stars, 1,060,385 ratings

Already installed, free, on-device, searchable. Start here and only leave if something specific stops you.

Just Press Record — 4.3 stars, 1,520 ratings

A genuine one-time purchase: $6.99 on the US App Store as of 11 August 2026. Not a subscription. In 2026 that alone is a feature.

The reason people buy it is the Apple Watch complication: one tap on your wrist starts recording, no phone involved, and it syncs through iCloud. If your ideas arrive while your hands are full, that’s the pitch.

Note the rating: 4.3 across 1,520 ratings is the lowest star average in the “capture” group here. Read recent reviews before you buy, since it’s a paid app with no free trial.

Voicenotes — 4.8 stars, 6,589 ratings

Built around the idea that the recording is raw material: it transcribes, then generates summaries and lists from what you said. Cloud-processed.

AudioPen — 4.7 stars, 143 ratings

Same premise, tighter scope: you ramble, it returns cleaned-up prose. Very small rating count, so treat it as promising rather than proven.

If you want voice as an input method

Wispr Flow — 4.8 stars, 13,308 ratings

Not a voice notes app. It’s a keyboard: you speak, and text appears wherever you were typing. Different job, and often the one people actually wanted when they went looking for a voice notes app.

Superwhisper — 4.4 stars, 808 ratings

Dictation with an on-device model option, from SuperUltra, Inc.

The privacy question, and how to test it

Every “on-device” claim in this article comes from a developer’s own store listing. You don’t have to take that on faith.

Turn on airplane mode and record something. If the transcript appears with no connection, it ran locally. If it hangs, spins or errors, your audio was going somewhere.

That thirty-second test is worth more than any privacy policy summary, because it checks the version of the app you installed rather than the one someone reviewed two years ago.

Apple’s Voice Memos and Notes are the safe assumption. Otter, Rev and Notta upload by design — that’s not a criticism, it’s the architecture that lets them do speaker labelling well.

The awkward questions

How accurate is any of this, really? Nobody will give you a straight number, including me, because accuracy depends almost entirely on your accent, your microphone and your background noise. A quiet room with a close mic gets near-perfect results from most of these. A windy street gets mush from all of them.

Test with your own voice on your own worst-case recording before you pay for anything. Every app here that costs money has either a free tier or a trial.

Can I get my transcripts out? Text export exists everywhere. What generally does not survive a move is the link between the transcript and the audio, plus any tags or folders. If you care about keeping recordings long-term, export to plain text into a notes system you control rather than trusting an app’s library.

What about languages other than English? Apple’s Notes transcription lists a specific set of supported device languages, and Apple states Voice Memos transcription isn’t available in all regions. Whisper’s language support is documented in OpenAI’s repository rather than as a marketing number. If you work in a language other than English, check the specific app’s own list rather than trusting a round number in a blog post.

Do I even need a dedicated app? Often not. Voice Memos plus a weekly habit of skimming transcripts beats an expensive app you open twice.

What to actually do with voice notes

The app is the easy part. The reason most people’s voice notes are useless is that recording feels like progress and nothing happens afterwards.

  • Skim the transcripts weekly. Ten minutes. Most of it is noise, which is fine — the point is finding the two that aren’t.
  • Move anything good into your real notes system. A voice note is a capture buffer, not an archive.
  • Delete aggressively. A hundred untouched recordings makes the search results worse, not better.
  • Say the context out loud. “This is about the pricing page” at the start of a recording makes it findable later. Your future self has no idea what “so, the thing” refers to.

FAQ

Which voice app transcribes best? For multi-speaker meetings, Otter, because speaker labelling is the feature nothing else on this list does as well. For notes to yourself, Apple’s Voice Memos is accurate enough, free, and processes on-device.

Is Whisper a free app I can download? No. Whisper is an open-source model from OpenAI released under the MIT licence. The consumer apps built on it — Whisper Transcription, Aiko, Superwhisper — are separate commercial products with their own pricing.

Is there a voice notes app with a one-time purchase? Just Press Record, $6.99 on the US App Store as of 11 August 2026, and Aiko at $24.00. Almost everything else is subscription-based.

Does Apple transcribe voice recordings for free? Yes, in both Voice Memos and Notes, on-device. Apple documents feature requirements and regional limits in its iPhone user guide.

Can I transcribe recordings I already have? Voice Memos transcribes recordings made in earlier versions automatically. Most third-party apps accept imported audio files. Check before you commit to a subscription for a backlog job.

Do I need internet? Not for Apple’s built-in transcription or for the on-device modes in the Whisper-based apps. Yes for Otter, Rev and Notta.

Sources

Related reads:

— Dolce