A level up on voice interaction with agents – and local

danielfromtas1 pts1 comments

Narrato | On-device dictation & read-aloud for Mac

Dictate and read aloud, in every app.<br>On-device speech-to-text and natural read-aloud, working wherever there’s text on your Mac. Your audio and text never leave the device.<br>Try Narrato free$29.99 USDWatch the 32-sec demo →<br>7-DAY TRIAL/PAY ONCE/NO ACCOUNT

Inserted00:00<br>Do anything<br>Full access<br>5.5 High

NarratoWork locallymain

Engineering<br>Runtime design<br>Audio graph<br>Model loading<br>Daily<br>AI Work<br>2026-06-12<br>2026-06-13<br>Releases<br>Research<br>Roadmap<br>Notes<br>Pasted image 20260628PNG<br>Pasted image 20260629PNG

Engineering / Runtime design<br>Runtime design notes<br>The runtime keeps both models resident once they are warm, so neither has to reload between sessions. Dictation and read-aloud share one audio graph: the recorder hands frames to the transcriber as they arrive, and the synthesizer streams the first chunk of a long passage while the rest is still being prepared. Everything stays inside a single process, which keeps the pipeline easy to reason about and simple to keep on-device.

0 backlinks1,767 words12,319 characters

New Message<br>To:AAlex Rivera<br>Subject:Build status

Write your message…<br>Send

NNarrato<br>Recents<br>Pricing<br>Press kit<br>Walking & ideas<br>Roadmap<br>More

Why walking helps us thinkShare<br>Why walking helps us think<br>At a desk, a difficult problem can shrink until it fills the whole screen. Walking changes the scale. The body settles into a rhythm, attention loosens, and ideas that felt unrelated begin to connect. You are not escaping the work; you are giving the mind enough distance to reorganise it. That is why the answer often arrives halfway around the block, when you have stopped trying to force it.

REAL WORKFLOW · 32 SECONDS<br>Talk through the problem. Narrato brings the context.<br>Dictate feedback, capture screenshots as you speak, and drop the complete brief into Codex.<br>Your browser does not support embedded video.[ NARRATO · CLIPBOARD CONTEXT ]

THE SPEC<br>Everything, on one line each.<br>DICTATION<br>Speech to textLands at your cursor, in any app<br>VocabularyLearns your terms, locally<br>Clipboard historyRecent dictations and copies<br>LanguagesDictate in 25 languages

Explore dictation →<br>READ ALOUD<br>Voices54 voices, 9 languages, all Kokoro↗<br>SpeedAdjustable, slow to fast<br>SourceAny selectable text on screen

Hear the voices →

100% on-device, no accountGlobal keyboard shortcutsWorks fully offlineVerify it →

Clipboard ContextContext is inserted automatically from your clipboard.

On the dashboard, these chart bars are way too thin screenshot, make them bolder. This heading copied text should fit on one line. And the metric cards screenshotdon't line up, so match their spacing.<br>See how dictation works →

0:04<br>0:09<br>0:13

Copy anything while you talk. It lands here, at the exact word.

localhost:5173

Weekly active users18,204 ▲ 6%

MONTUEWEDTHUFRISATSUN

Copied to clipboard at 0:04 , inserted where you said it.

Copied to clipboard · 0:09 Total revenue generated across every connected workspace and billing account for the current month.

The exact text you copied, dropped in as raw context.

Dashboard×<br>app.dashly.io/overview<br>Users12,480▲ 4.2%<br>Revenue$8,210▲ 9.1%<br>Churn1.2%▼ 0.3%

Copied to clipboard at 0:13 , a different app, same drop-in.

READ ALOUD · VOICES<br>Find a voice you’ll actually use.<br>Every voice is synthesized locally with Kokoro. Audition a few here; the full library spans nine languages.<br>ENGLISH · UK<br>Daniel

0:000:07

American & British English, Spanish, French, Italian, Portuguese, Hindi, Japanese & Mandarin. 54 voices across 9 languages.

01DanielENGLISH · UK02HeartENGLISH · US03AdamENGLISH · US04EmmaENGLISH · UK05DoraSPANISH · ES

Hear all 54 voices · 9 languages →

FREQUENTLY ASKED QUESTIONS<br>Questions, answered.<br>Everything a careful buyer asks before downloading a beta. Tap any question.<br>Is it really 100% on-device?Yes. Transcription and narration run entirely on your Mac, so your audio and text never leave the device, and everything works offline. The beta does make a few network calls that never carry your content: anonymous, content-free diagnostics and crash reports (both switch-off-able in Settings), and a one-time download of the dictation and voice models on first launch.

How does the free trial work?Download the notarized DMG and use every feature free for 7 days, no card required. After that, a one-time license unlocks it for good.

What do I get for $29.99 USD?A perpetual license for 2 Macs, every beta update plus 1.0 when it lands, and a 14-day no-questions refund. It’s the beta price, and it locks in for early supporters. Pay once. There’s no subscription, ever.

Why pay once when everything else is a subscription?Because your Mac does the work. Both models run on your own silicon, so there is no server bill for us to pass on, and no usage caps to meter. We'd rather earn the next version than rent you this one.

What are the system requirements?macOS 14 or later on an Apple Silicon Mac.

Why isn’t it on the Mac App Store?The App Store sandbox blocks the...

device text clipboard dictation read aloud

Related Articles