// VOICE DICTATION (ON-DEVICE SPEECH-TO-TEXT) REVIEW

Yap Review (2026): Honest Verdict on the Free, On-Device macOS Dictation App With No Model to Download

Yap review (2026): honest verdict on Frigade's free, on-device macOS voice dictation app that needs no model download — accuracy, privacy, limits, and value.

KompozyTurn one idea into a week of content — across every platform, published for you.
Get Started →
Last verified · 2026-07-28 · by Moe Ameen
The verdict
4.2 / 5

Yap is one of the cleanest on-device dictation tools on the Mac: free, MIT-licensed, native Swift, and — its headline trick — it downloads no speech model, leaning on Apple's built-in macOS 26 models instead. For private, fast, everyday dictation it's an easy recommendation. The limits are honest ones: it needs macOS 26 and runs on Mac only, language follows your system locale, and it does exactly one thing — get speech to text at your cursor. If you want dictation, it's a strong yes. If you wanted the tool that makes and publishes content from what you say, that's a different category.

Yap is a free, open-source voice dictation app for macOS from Frigade, shared as a "Show HN" post in late July 2026 with the pitch "OSS on-device voice dictation for macOS with no model to download." The workflow is the classic one: press a global shortcut, speak, press again, and your words are typed into whatever field you were in. What makes it worth a review is the design choice under the hood — it ships no speech model of its own.

Instead of bundling Whisper or downloading a multi-gigabyte model, Yap calls Apple's native SpeechAnalyzer and SpeechTranscriber APIs introduced in macOS 26 (with SFSpeechRecognizer as a fallback), so it transcribes with models the OS already carries. The payoff is a roughly 4 MB app, around 3,000 lines of native Swift, that idles near 60 MB of memory and never touches the network — no account, no API key, no cloud. That's a genuinely different philosophy from the paid, model-heavy dictation tools it's reacting to.

This review scores Yap on what it actually is: a focused, private dictation utility. It does not penalize Yap for lacking content generation or publishing, because it never claimed those — but it does score platform reach and language flexibility honestly, since those are real constraints for real users. Ratings and details below reflect the app as of 2026-07-28.

What Yap is

Yap is a menu-bar dictation app for macOS. You bind a global shortcut (default ⌘⇧D), speak, and it inserts the transcript into the active app by placing the text on the clipboard, issuing a paste, and restoring your prior clipboard contents. Audio is captured with AVAudioEngine and transcribed on-device via Apple's macOS 26 Speech APIs, so nothing leaves your Mac. It's MIT-licensed and free, installable via a Homebrew cask (brew install --cask frigadehq/tap/yap) or a direct .dmg. The feature set is deliberately small and well-chosen: a rebindable shortcut, a live waveform while you speak, a local searchable transcript history stored with SwiftData (copy or delete past entries), single-modifier trigger support, microphone auto-detection, and an optional launch-at-login toggle. The maker reports Apple's on-device model landing slightly ahead of Whisper Small on accuracy and roughly three times faster, which matches the everyday experience even if it isn't an independent benchmark. Note that several unrelated open-source projects also go by "Yap" — this review is Frigade's app specifically.

Who Yap is for

Yap fits Mac users on macOS 26 who want fast, private, no-friction dictation and don't want to pay for it or manage a downloaded model. That's writers, developers, students, and anyone who thinks faster than they type and wants speech-to-text that works in every app. It's an especially good fit for the privacy-conscious, since audio never leaves the machine. It's a poor fit for anyone on Windows, Linux, or an older Mac (it simply won't run), for people who need on-demand multi-language dictation, and for anyone hoping a dictation app will also produce or publish content — that's a job for a content engine, not a transcription tool.

Scoring breakdown

DimensionScoreWhy
Transcription accuracy4.3 / 5Apple's on-device model is solid for everyday speech; the maker reports it slightly ahead of Whisper Small, though that's not independently benchmarked.
Speed & latency4.6 / 5On-device transcription is fast, with the author citing roughly 3x the speed of Whisper Small.
Privacy5.0 / 5Fully offline — no account, no API key, no network call, audio never leaves the Mac.
Ease of use4.5 / 5Bind a shortcut, speak, done; the live waveform and searchable history are thoughtful touches.
Footprint & performance4.8 / 5About 4 MB and ~60 MB idle memory as native Swift — far lighter than bundled-model or Electron rivals.
Platform support2.5 / 5Requires macOS 26 and runs on Mac only; Windows, Linux, and older-Mac users are shut out entirely.
Language flexibility3.0 / 5Follows the system locale, with a language picker only noted as a likely future addition — no on-demand multi-language dictation yet.
Value5.0 / 5Free and MIT-licensed with no subscription; for what it does, the price is unbeatable.
Feature depth3.5 / 5Focused by design — no custom vocabulary, text commands, or engine choice; it does dictation and stops there.

Pros and cons

Pros

  • Completely free and MIT-licensed — no account, API key, or subscription.
  • Fully on-device and offline, so audio never leaves your Mac — a real privacy advantage over cloud dictation.
  • No model to download: it uses Apple's built-in macOS 26 speech models, keeping the app tiny (~4 MB) and light.
  • Native Swift rather than an Electron wrapper, so it's fast and unobtrusive.
  • Clean, focused UX: rebindable shortcut, live waveform, and a local searchable transcript history.
  • Fast enough for real work — the maker reports about 3x the speed of Whisper Small.

Cons

  • Requires macOS 26 (Tahoe) and runs on Mac only; no Windows, Linux, or older-Mac support.
  • Dictation is the whole product — no content generation, repurposing, scheduling, or publishing.
  • Language follows the system locale with no on-demand multi-language option yet.
  • You can't swap the speech engine; accuracy is whatever Apple's on-device model delivers.
  • No custom vocabulary or text commands, so technical jargon and names may need correcting.
  • Young project sharing its name with several unrelated tools, so support and discovery are thin.

Pricing analysis

There's little to analyze on price because Yap is free and MIT-licensed. There's no tier, no trial, no upsell — you install it via Homebrew or a downloaded .dmg and use it. Your only real cost is owning a Mac that runs macOS 26, since the app depends on those newer Speech APIs. That makes it dramatically cheaper than paid dictation tools like Wispr Flow or the many subscription "AI voice" apps, and cheaper in resource terms than free tools that bundle a local Whisper model you have to download and store.

The trade you're making for free is scope and lock-in to Apple's stack. Yap gives you exactly one capability — private, local speech-to-text — and none of the extras that paid tools justify their subscriptions with (custom vocabularies, cross-platform sync, AI rewriting, or content workflows). For someone who only wants dictation, that's a great deal and the absence of a price tag is a feature. For someone who expected dictation to be a stepping stone to producing content, the honest read is that no price would make Yap the wrong tool right — it's simply not built for that, and the thing that is (an engine that generates and publishes) is a separate purchase.

Use-case fit

Use caseFitWhy
Everyday dictation across Mac appsStrongThis is exactly what Yap is for — a shortcut, your voice, and text at the cursor in any field.
Privacy-sensitive transcriptionStrongFully offline and on-device, so no recording or transcript is ever sent anywhere.
Hands-free idea and note captureStrongSpeak a thought the moment it lands and have it saved to a searchable local history.
Coding comments and technical dictationOKWorks fine, but with no custom vocabulary, jargon and product names may need cleanup.
On-demand multi-language dictationWeakLanguage follows the system locale; there's no in-app language picker yet.
Windows, Linux, or older-Mac usersWeakYap requires macOS 26 and runs on Mac only, so it simply won't run for these users.
Turning what you say into published contentWeakYap ends at a transcript; making and scheduling video, carousels, and posts is a content engine's job, not a dictation app's.

Alternatives worth considering

  • Superwhisper — a Mac dictation app built around downloadable local Whisper models, with more configuration if you want to pick and tune the engine.
  • Wispr Flow — a polished, paid cross-platform dictation tool with AI cleanup, for users who want a supported commercial product across OSes.
  • Apple Dictation — the built-in system dictation, already free on every Mac and iPhone, if you don't need a dedicated shortcut app or local history.
  • MacWhisper — a well-known Mac app for transcribing files and dictation using local Whisper models, stronger for batch/file transcription.
  • Kompozy — not a dictation app but the layer after it: it turns a dictated draft into video, carousels, images, a blog, a newsletter, and posts, then publishes across platforms.

How Kompozy compares

Being straight about it: Kompozy is not a Yap alternative, and if you came here to pick a dictation app, Yap is a better answer than Kompozy for that specific job. Kompozy doesn't do speech-to-text at all. Where it fits is the step Yap can't reach — after the transcript. A common reason people evaluate dictation tools is that they want to make content faster, and dictation only solves the "getting words down" fraction of that. Once the words exist, Kompozy takes over: paste a dictated script or angle in as a source and it generates Persona Shorts and HeyGen avatar video, Clipped Shorts, brand-exact Carousels, Photo Posts, Quote Graphics, a Blog Article, an Email Newsletter, and Text Posts, all held to your voice by a Persona Brief, then schedules and publishes them across eight social platforms plus blog and email. So the honest recommendation is to use both for what each is good at — Yap to capture your voice privately and for free, Kompozy to turn that captured voice into finished, published content — rather than expecting either one to do the other's job.

Frequently asked questions

Is Yap worth it?

For Mac users on macOS 26 who want fast, private dictation, yes — it's free, MIT-licensed, tiny, and downloads no model because it uses Apple's built-in speech engine. The main reasons it wouldn't be worth it are simple: you're not on a supported Mac, you need multi-language dictation, or you actually wanted a tool that makes and publishes content, which dictation isn't.

Why does Yap not need a model download?

Yap uses Apple's native SpeechAnalyzer and SpeechTranscriber APIs introduced in macOS 26 (with SFSpeechRecognizer as a fallback), so it transcribes with the models the operating system already ships. That's why the app is only about 4 MB and there's nothing multi-gigabyte to download, unlike tools that bundle Whisper.

What are the requirements for Yap?

Yap requires macOS 26 (Tahoe) or later because it depends on those newer Speech APIs, and it runs on Mac only — there's no Windows or Linux version. It's installable via a Homebrew cask or a direct .dmg download, and it needs microphone access.

How accurate is Yap's transcription?

Accuracy comes from Apple's on-device model, which is solid for everyday speech. The maker reports it landing slightly ahead of Whisper Small on accuracy and roughly 3x faster, though that's a developer claim rather than an independent benchmark. Because there's no custom vocabulary, unusual names and jargon may need occasional correction.

Is Yap free?

Yes. Yap is free and MIT-licensed, with no account, subscription, or API key. Your only cost is a Mac running macOS 26. That makes it cheaper than paid dictation tools and lighter than free ones that download and store a local Whisper model.

Can Yap create or post content?

No. Yap only transcribes speech into text at your cursor; it doesn't generate video, images, carousels, or posts, and it can't schedule or publish. If that's what you need, pair Yap with a content engine like Kompozy, which turns a dictated draft into finished multi-format content and publishes it across platforms.

Yap vs Superwhisper — what's the difference?

Both are Mac dictation apps, but Yap uses Apple's built-in macOS 26 speech models so it downloads nothing and stays tiny, while Superwhisper is built around downloadable local Whisper models with more engine configuration. Yap is simpler and lighter; Superwhisper gives you more control over the model.

Related deep guides

See Yap vs Kompozy comparison → · Get Started →