I wanted voice dictation on macOS to feel instant and native, and I built Talkify an 8.2 MB macOS dictation app with 123 ms median release-to-text latency. It was the fastest of five apps tested ahead of superwhisper, voiceink, wispr flow and macwhisper.

It lives in your notch and have beautiful shader animations and different kind of customizations.

Talkify is built on top of Apple’s SpeechAnalyzer using system-managed speech models to transcribe audio entirely on-device, its free and open source.

https://usetalkify.app https://github.com/tornikegomareli/Talkify…

Talkify - Speak and it’s written

Source: https://usetalkify.app/

Powered by Apple’s new SpeechAnalyzer.

Hold your key and Talkify hands your voice to the speech engine Apple introduced in macOS 26, the same one behind Notes, Voice Memos and Journal. Let go and the finished text goes straight into whatever you were typing in.

It answers as you speak

The model was built for live transcription that needs low latency without giving up accuracy, so words arrive while you are still talking rather than after you stop. Talkify warms it up at launch, so your first syllable meets an engine that is already running. There is no spin-up, and no server to wait for.

Built for real speech

Apple’s new model is faster and more flexible than the one dictation used before, and it holds up on long, conversational speech, even when you are not close to the microphone. Talkify shows its first rough guess immediately, then replaces it as the recognizer settles on the final wording.

Nothing to install or manage

The language model lives in system storage, downloaded and kept up to date by macOS. It doesn’t inflate Talkify’s size or memory, it runs outside the app entirely, and your speech never leaves the Mac to reach it.

It lives in the notch.

Talkify turns the notch into the one place your Mac listens from and gives it a voice visual worth watching.

Waveform

Edge Glow

Compact

Stop talking. It’s already there.

123milliseconds from letting go of the key to your words in the input. No spinner, no “processing”, no server to wait for.

Dictation speed

Every engine was tested on the same machine: a MacBook Pro (M4 Max, 14-core, 36 GB) running macOS 26.5.

TalkifyApple SpeechAnalyzer

123ms

superwhisperStandard EN, local

216ms1.8×

VoiceInkParakeet V3, local

324ms2.6×

Wispr FlowDefault, cloud

587ms4.8×

MacWhisperLarge v3 Turbo, local

1,149ms9.4×

Every engine spoke the same phrase into the same editor, so recognition and text insertion are both inside these numbers.The harness and the full dataare public. Rerun it yourself.

Your transcriptions stay on your Mac.

**No account.**Nothing to sign up for, nothing to sign into. Download it and dictate.

**No internet.**Recognition runs on the Neural Engine with Apple’s on-device models. Talkify makes zero network requests, so airplane mode changes nothing.

**Stored on your disk only.**Your text is written to the app you were typing in and nowhere else. Insights are computed and kept locally, as a plain file on your own disk.

8.9 MBto download11.5 MBinstalledTalkify is small because it does not carry a speech model: macOS keeps one and shares it with every app, so nothing about recognition lives inside this download. No audio and no transcripts are ever written to disk.

Two languages. Two keys.

Pick a second language and it gets its own key. Hold one for English, the other for German, with no setting to change in between.

Nothing to switch

Both languages stay loaded, so the second answers as fast as the first, and the notch shows which one is listening. Thirty locales are available across German, English, Spanish, French, Italian, Japanese, Korean, Portuguese, Cantonese and Chinese. A language you haven’t used before downloads its model once.

fnEnglish ·right ⌥German

It never guesses

Apple’s recognizer transcribes one language at a time and cannot tell you which language it just heard. Detecting it would mean transcribing first and inferring from the result, and in the wrong language that result is fluent nonsense rather than an error. A key you press is instant and always right.

It reads back, and it’s yours to tune.

Read Aloud

Select text in any app and Talkify speaks it, with Apple’s Enhanced and Premium voices, or with your own Personal Voice once you authorize it. Voices are listed with their language and quality, the list refreshes the moment a new one finishes downloading, and playback runs on device and offline.

⌥escRead Aloud

Made to your taste

Three voice visuals, nine waveform styles, six glow palettes, session sounds, and rebindable shortcuts, every change previewed live before you dictate. Insights then tells you what you actually said: words per day, pace, voice time, streaks.

Questions

What is Talkify?

Talkify is a free, open-source dictation app for macOS. You hold a key, speak, and release, and your words are typed into whatever app you were using. It lives in the menu bar and shows a live voice visual in the MacBook notch while it listens.

How fast is Talkify?

About 123 milliseconds from releasing the key to your words appearing in the input, measured on a MacBook Pro M4 Max. Running the same phrase through the same editor, superwhisper took 216 ms, VoiceInk 324 ms, Wispr Flow 587 ms and MacWhisper 1149 ms. Recognition and text insertion are both inside those numbers. The harness and full data are public at github.com/tornikegomareli/talkify-latency-bench.

Is Talkify more accurate than other dictation apps?

Not necessarily. Accuracy was comparable across engines in the same benchmark, and Talkify was the least consistent of them: it finalizes soonest, so it commits with the least trailing audio. Talkify’s advantage is speed, not accuracy, and version 1 inserts raw text with no cleanup pass.

How much space does Talkify use?

The download is 8.9 MB and the installed app is 11.5 MB. It stores about 20 KB of its own data: word counts and dates for the local Insights screen. No audio and no transcripts are written to disk. Talkify carries no speech model of its own, because macOS keeps one and shares it with every app that asks.

Does Talkify work offline?

Yes. Recognition runs on device using Apple’s SpeechAnalyzer, the speech engine introduced in macOS 26. Talkify makes no network requests at all, so it works in airplane mode and nothing you say leaves your Mac.

Is Talkify free?

Yes. Talkify is free and open source, with no account, no subscription and no trial. You can install it with Homebrew or download the DMG from GitHub.

What does Talkify require?

macOS 26 (Tahoe) on an Apple Silicon Mac. On first launch it asks for Microphone, Speech Recognition, Accessibility and Input Monitoring permissions, because dictation listens for a global key and inserts text into other apps.

How do I install Talkify?

With Homebrew, run: brew tap tornikegomareli/talkify https://github.com/tornikegomareli/Talkify, then brew install --cask talkify. You can also download Talkify.dmg from the latest GitHub release.

Is Talkify a Wispr Flow alternative?

It covers the same job (hold a key, speak, and have text appear in any app) but Talkify is free, open source, and runs entirely on device with no account and no cloud service.

Where does Talkify store my transcriptions?

Nowhere. Your text goes into the app you were typing in and is not saved by Talkify. The only thing kept on disk is aggregate usage data for the local Insights screen: word counts, speaking time and dates, never the words themselves.

Can Talkify dictate in languages other than English?

Yes. Thirty locales are available across German, English, Spanish, French, Italian, Japanese, Korean, Portuguese, Cantonese and Chinese, using the languages macOS itself supports. You can also set up a second language with its own key: hold one key for English, another for German, with nothing to switch in between. Talkify does not try to detect the spoken language, because Apple’s recognizer transcribes one language at a time and a wrong guess produces fluent nonsense instead of an error.

Can Talkify read text out loud?

Yes. Select text in any app and press the Read Aloud shortcut, and Talkify speaks it with an Apple Enhanced or Premium voice, or with your own Personal Voice once you authorize it. Playback is on device and works offline.

Read it, fork it, or just star it.

Talkify is free because it costs nothing to run. Your Mac does the work, and the whole thing is on GitHub: no telemetry to audit, no server to trust, nothing hidden.

If you want more of this, a star is the cheapest way to say so. It’s how the next person finds it.

Talk to your Mac.

You already know how to use it. You’ve been practicing since you learned to speak.

Free & open source · macOS 26 · Apple Silicon