What's new.

Polish styles became a full system. Five built-in styles — Verbatim (no AI), Minimal, Shape, Formal, and the new Coding — each bindable to its own hotkey, so different keys speak different registers. Shape was rebuilt as the flagship: it boldly reorganizes rambling speech while keeping your voice and intent. Translate styles turn dictation into instant translation: speak one language, land another. Preferences were reorganized around plain language throughout.

Smart Routing matured into one of Acousmos's defining features: one trigger key, a different style per app — literal in your IDE and terminal, casual in chat, formal in email — with optional per-app engine overrides. The Playground landed alongside it: compare the same dictation across styles and AI engines side by side, and keep what wins. Reliability hardened too: failed transcriptions recover from saved audio with one-click retry, and your plan and remaining trial words show right in the menu bar.

Acousmos launched commercially: Plus and Pro plans with website checkout, a manage-subscription portal, and one-click license activation via deep link — no account system, ever. Every new install starts with a one-month free Pro trial, no signup, no card. The app moved to industry-standard Sparkle auto-updates, and managed-cloud recognition began routing per region across multiple engines, keeping dictation fast wherever you are.

Recognition expanded across 100+ languages, with mixed-language speech handled natively — say a Chinese sentence with English terms in it, and both land correctly. A quick language switcher in the menu bar made mid-day switching one click, and translation gained a proper target-language setting. Input-method state is preserved on every insertion, so dictating never disturbs how you type.

The three-layer dictionary — vocabulary biasing, smart corrections, exact replacements — began learning from your edits: fix a name once and it proposes the correction forever after. Snippets made frequent phrases a single utterance. Stats put dictation volume on a chart, and Storage controls made the archive's footprint visible and manageable.

The core loop that still defines Acousmos: press a hotkey in any macOS app, speak naturally, and polished text lands at your cursor — with AI clean-up styles from day one. And from the very first build, the archive was the point: every dictation kept locally, original audio and text together, full-text searchable in English and Chinese. Toggle and push-to-talk modes, custom vocabulary, and a guided permissions setup rounded out the start.