Best dictation apps for Mac in 2026

Most lists of the best dictation apps for Mac rank them by feature count. That misses the two questions that actually change the decision: where your speech gets turned into text, and whether you pay once or every month.

Both are easy to check and hard to argue with, so this list is organised around them.

How these were picked

Every app below meets three requirements: it runs on macOS, it types into other applications rather than only into its own window, and its maker states its price publicly.

Each entry is judged on five things:

Where recognition happens. On your Mac, or on someone else’s computer. Capturing audio locally and recognising it locally are different claims, and only the second one means the recording never leaves.

How you pay. One-time licence, subscription, or free.

What else it does besides live dictation, such as transcribing files or recording meetings.

Hardware requirements, which in 2026 mostly means whether Intel Macs are still supported.

What it is bad at. Every entry names something, including ours.

Prices and claims were read from each maker’s own site on 19 August 2026. Ignore any number here that contradicts the vendor’s current page, and tell us so we can fix it.

Free, and already on your Mac

Apple Dictation

macOS has dictation built in, and most people never switch it on. It costs nothing, works in every text field, and on Apple silicon Macs Apple documents that voice inputs and transcripts for general text dictation are processed on the device rather than sent to Siri servers.

Turn it on under System Settings, Keyboard, Dictation. There is no cap on how long you can talk; it stops after 30 seconds of silence.

What it is bad at: there is no vocabulary list, so names, acronyms and technical terms get corrected by hand forever. There are no formatting rules, no transcription of files, and no way to tune anything beyond language and shortcut.

Worth trying first. If it covers your needs, everything below is a waste of money. We went through what it can and cannot do in more detail in voice to text on a Mac.

Spokenly

Free, and more capable than free usually means. Its site lists unlimited local transcription on the free tier, using Whisper and Parakeet models that run offline on Apple silicon, plus bring-your-own-key cloud transcription. A Pro tier at $9.99 a month adds managed cloud models and priority support.

The detail that matters to a particular audience: it ships an MCP server for AI coding agents such as Claude Code and Cursor. If you want an agent to transcribe audio through a local tool instead of a metered API, this is one of only two options here that offers it.

What it is bad at: a free tier funded by an optional subscription is a business model with a question mark over it, and support expectations should be set accordingly.

Pay once

VoiceInk

$29 for the Solo licence, $49 and $69 for wider ones, all lifetime. Its site states that all voice transcription is processed locally on your device and that your voice data never leaves your Mac. The source is public on GitHub, so that claim is checkable rather than promised, which is rare in this category and worth more than most feature lists.

Requires Apple silicon and macOS 14.4 or later, with 8 GB of RAM recommended. There is a free trial.

It is dictation and vocabulary, not a transcription suite: the site describes no file import, no meeting recording and no MCP server.

What it is bad at: no Intel support, and reading source code is only reassuring to people who can read source code. That said, if an auditable codebase matters to you more than anything else on this page, this is the one to take.

MacWhisper

€64 once, including lifetime updates, with a free tier that already covers recording, common audio and video formats, 100 languages, export to SRT, VTT and TXT, and system-wide dictation. Its site states that local models do the transcription, so sensitive files are processed without data ever leaving your Mac, and lists both Whisper and Parakeet among supported models.

It is the strongest of the group at files rather than live speech: drag in a recording, search transcripts, edit them properly, record meetings from Zoom, Teams and others. If files are the whole reason you are here, the routes are compared in how to transcribe audio and video to text on a Mac.

What it is bad at: it is the most expensive one-time licence here, and if you only ever dictate short messages you are paying for an editor you will not open.

Kekoso

Ours, so weigh the description accordingly. $29 once, a 7-day trial, up to 3 machines, and every part of it runs on your Mac: dictation into any app, audio and video files, YouTube links, meeting recordings with the microphone and system audio on separate tracks, a vocabulary you fill in yourself, and a local MCP server your coding agent can call. Recognition happens on the Neural Engine, and the audio is discarded once the text lands.

That list is the reason it exists. Covering all of it today usually takes a dictation app, a transcription tool and a meeting service, most of them subscriptions, and most of them recognising your speech somewhere other than your desk. Three subscriptions at $8 to $15 a month each is $300 to $500 a year, every year.

What it is bad at: it needs macOS 14 or later on Apple silicon, so every Intel Mac is out, and there is no Windows or Android version to switch to. It is closed source, so the claim about locality rests on our word rather than on a repository you can read. And it is the newest product on this page, without the multi-year track record the others have earned.

Subscriptions

superwhisper

A free tier covering voice-to-text in any app, meeting recording, 100+ languages and unlimited small AI models. Pro is $8 a month, with a yearly option described as two months free, and adds your own API keys, unlimited cloud and local models, translation, and transcription of audio and video files. A lifetime licence exists but the pricing page does not print its price.

It works offline, and it is the only app here that also runs on Windows and iOS. Its own site notes that offline models only run really well on Apple silicon and that Intel Macs work best with cloud models, which is a useful thing to have in writing before you buy.

What it is bad at: the model selection is a project in itself. The flexibility that makes it powerful is the same thing that makes the first hour confusing.

Wispr Flow

$12 per user a month billed annually, or $15 monthly. The free tier allows 2,000 dictations a week on desktop, 1,000 on iPhone, unlimited on Android, and includes a notetaker with speaker identification and MCP integration for Claude and ChatGPT.

It has the widest reach of anything here: desktop, iPhone and Android, with the mobile tiers more generous than the desktop one.

What it is bad at, and this is the important one: the pricing page says nothing about where recognition happens. For most people that does not matter. If you are under NDA, or you handle client material, the absence of a statement is itself the answer, and $144 to $180 a year buys you a service rather than a tool you own.

Aqua Voice

$8 a month for unlimited words, $24 a month for a Max tier adding realtime mode and voice commands, $12 per user for teams. The free Starter tier gives you 1,000 words in total and five custom dictionary entries, which is enough to evaluate it and not enough to use it.

Its site describes a Privacy Mode where nothing is stored on their servers at all, but does not state where recognition itself happens by default.

What it is bad at: a 1,000-word free tier is a demo rather than a trial, and the gap between $8 and $24 tiers puts the features people quote in reviews behind the higher one.

The table

Read from each maker’s own site, 19 August 2026.

App	Price	Recognition	Files	Meetings	MCP	Platforms
Apple Dictation	Free	On device, Apple silicon	No	No	No	macOS
Spokenly	Free, Pro $9.99/mo	Local models offline	Not stated	Not stated	Yes	macOS, iOS, Windows, Linux
VoiceInk	$29 once	On device	Not stated	Not stated	Not stated	Apple silicon, macOS 14.4+
MacWhisper	€64 once	Local models	Yes	Yes	Not stated	macOS
Kekoso	$29 once	On device	Yes, plus YouTube links	Yes, separate tracks	Yes, local	Apple silicon, macOS 14+
superwhisper	Free, Pro $8/mo	Local or cloud	Yes, on Pro	Yes	Not stated	macOS, Windows, iOS
Wispr Flow	$12 to $15/mo	Not stated	Notetaker	Yes	Yes	macOS, iOS, Android
Aqua Voice	Free 1,000 words, $8/mo	Not stated	Not stated	Not stated	Not stated	macOS, iOS, Windows

“Not stated” means exactly that: the maker does not say on its own site. It is not a claim that the feature is missing, only that we would not print a claim they never made.

What to pick

The best dictation app for you falls out of two answers: how often you dictate, and whether the recording is allowed to leave your machine.

If you have never dictated before, start with the one already installed. Apple Dictation costs nothing and answers the only question worth answering first: whether you like working this way at all.

If you want the best free speech to text app that is not built into the system, Spokenly runs locally, costs nothing, and has the MCP server.

If you dictate most days and dislike subscriptions, the arithmetic is not subtle. VoiceInk and Kekoso are both $29 once, which is less in year one than any subscription here costs in five months. Between the two: VoiceInk if you want dictation with an auditable codebase, Kekoso if you also want files, YouTube links, meetings and an MCP server without adding a second and third tool.

If you currently pay for a dictation app, a transcription tool and a meeting notetaker separately, that stack is the case for Kekoso: the same four jobs, one purchase, and nothing uploaded to anyone.

If your work is mostly recordings rather than live speech, MacWhisper. The editor and the meeting recording are what you are paying €64 for.

If you switch between a Mac, a PC and a phone, superwhisper is the only one here that covers all three.

If none of that applies and you just want the smoothest possible experience with a company that will keep polishing it, Wispr Flow is a good product at a subscription price. Just decide first whether an unstated recognition location is acceptable for the material you dictate.

Questions people ask

What is the best speech to text app for a Mac?

There is no single answer, because the apps differ on price model rather than on quality of recognition: several of them run the same underlying models. For occasional use, the built-in Dictation is free and already installed. For daily work without a subscription, VoiceInk and Kekoso are $29 one-time. For recordings rather than live speech, MacWhisper. For cross-platform use, superwhisper.

Is there a free dictation app for Mac?

Two, in fact. macOS has Dictation built in at no cost, and Spokenly offers unlimited local transcription on its free tier. Wispr Flow and Aqua Voice also have free tiers, but both cap how much you can dictate: 2,000 words a week on desktop for Wispr Flow, and 1,000 words in total for Aqua Voice.

Which Mac dictation apps work fully offline?

MacWhisper, VoiceInk and Kekoso recognise speech on your own machine. Spokenly does too on its free local models, and superwhisper can, though it also offers cloud models. Apple's built-in Dictation processes general text dictation on device on Apple silicon Macs. Wispr Flow and Aqua Voice do not state on their sites where recognition happens.

Should I buy a dictation app once or subscribe?

If you dictate most days, a one-time licence is cheaper within the first year: VoiceInk starts at $29 and Kekoso is $29, against $96 to $180 a year for the subscription apps. Subscriptions make more sense if you want the cloud models and the vendor's ongoing server-side features, since those cost the maker money every month you use them.

Do any of these work on Intel Macs?

superwhisper supports Intel Macs, though its own site notes that offline models only run really well on Apple silicon and Intel machines work best with cloud models. VoiceInk requires Apple silicon and macOS 14.4 or later, and Kekoso requires Apple silicon and macOS 14. Check the current requirements before buying if you are on an Intel Mac.

Which one is best for developers using AI agents?

Spokenly ships an MCP server for AI coding agents such as Claude Code and Cursor, and Wispr Flow lists MCP integration on its free tier. If you want an agent to be able to transcribe audio through a local tool rather than an API, those are the two to look at today.