Press a hotkey, speak in any Mac app, and clean text appears at your cursor — filler removed, grammar fixed, styles you set per app. Every word (audio + text) is saved to a private, searchable archive on your Mac. Local models and BYOK on Pro.
No reviews yetBe the first to leave a review for Acousmos
Maker
📌
Hi PH! Solo-building this for people who think out loud.
Three months of my own daily use: 11,669 dictations, 495k words, 188 WPM against my ~40 WPM typing — the app's stats page says 162 hours saved. But the number that changed my mind was retrieval: things I said weeks ago, found in seconds.
Every dictation tool I tried treats your words as disposable: speak, paste, gone. Acousmos makes the opposite bet — your voice is an asset. So besides fast dictation (hotkey → speak → polished text lands at your cursor in any app), everything you say is archived locally: original audio + text, full-text searchable in English and Chinese, open format, exportable, yours forever — subscription or not.
Things I sweated over:
• Per-app smart routing: assign a style to an app once — Coding for Claude Code/Cursor (terms stay exact, spoken rambling tightens into a clear prompt), formal in email — and the primary hotkey switches automatically from then on.
• CJK input methods: it switches your IME to paste and restores it after, so bilingual dictation doesn't wreck your typing state.
• Voice translation: set a target language once; speak Chinese, English lands at your cursor (or any pair). Same hotkey muscle memory.
• Privacy exits: the trial and Plus run on managed cloud (zero setup), but Pro can route through YOUR api keys or fully local models — airplane-mode dictation works.
• No account: download the DMG, first dictation starts a 1-month full Pro trial. No signup, no card.
Honest caveats: Apple Silicon Macs only (macOS 14+), no Windows/iOS, and we're pricier than some rivals ($12.99–17.99/mo — every plan covers 2 Macs) because the archive and managed cloud are the product, not an upsell. Public roadmap/feedback: https://acousmos.featurebase.app/
Try the engine with your own voice before installing: https://acousmos.com/arena/ — and I'd love brutal feedback on the archive concept. What would make a "voice second brain" actually useful to you?
Report
the per-app style memory is genuinely smart, like actually anticipating that my Slack voice differs from my notes voice. plus keeping everything local by default is a quiet flex most voice tools skip.
Report
Maker
@akkocaseng49172 Thank you, Şengül! The per-app routing came straight from my own annoyance — my Slack voice and my docs voice are different people, and toggling modes by hand never stuck. So you assign a style to an app once, and the hotkey just knows from then on.
And the archive never leaving your Mac is the core philosophy: AI can touch your words in transit, but the record of them belongs to you. If you give it a try, I'd love to hear what your archive turns up after a week.
Report
The per-app style settings are a really thoughtful touch, especially since most dictation tools just dump raw text wherever you are. Glad it all stays local too.
Report
Maker
@yamurobancsyyp Thanks Yağmur! The raw-text-dump thing is honestly what pushed me to build this — a terminal wants your exact words, an email wants full sentences, and most tools just don't care where the text lands.
One honest note on "all local": the archive does live only on your Mac, but the default transcription runs through managed cloud (transit only, nothing stored or trained on). If you want zero cloud, Pro runs fully local models — airplane mode and all.
the per-app style memory is genuinely smart, like actually anticipating that my Slack voice differs from my notes voice. plus keeping everything local by default is a quiet flex most voice tools skip.
@akkocaseng49172 Thank you, Şengül! The per-app routing came straight from my own annoyance — my Slack voice and my docs voice are different people, and toggling modes by hand never stuck. So you assign a style to an app once, and the hotkey just knows from then on.
And the archive never leaving your Mac is the core philosophy: AI can touch your words in transit, but the record of them belongs to you. If you give it a try, I'd love to hear what your archive turns up after a week.
The per-app style settings are a really thoughtful touch, especially since most dictation tools just dump raw text wherever you are. Glad it all stays local too.
@yamurobancsyyp Thanks Yağmur! The raw-text-dump thing is honestly what pushed me to build this — a terminal wants your exact words, an email wants full sentences, and most tools just don't care where the text lands.
One honest note on "all local": the archive does live only on your Mac, but the default transcription runs through managed cloud (transit only, nothing stored or trained on). If you want zero cloud, Pro runs fully local models — airplane mode and all.