Ten Surfaces, One Subscription

Short answer: install AICHE on whichever device is in front of you, sign in once, and the same voice-to-text layer runs on macOS, Windows, Linux, iPhone, iPad, Apple Watch, Android, Chrome, Obsidian, and a REST API. Pro covers up to 10 devices, and cloud sync is opt-in.

The problem this solves

Most voice apps make you pick an OS. Mac-only, or Mac + Windows, or "mobile but no desktop". If your work laptop is Windows, your personal machine is a Mac, your phone is Android, and your dev box runs Fedora, you end up with three different dictation products, three subscriptions, and three different output qualities. The voice layer becomes a function of which device you happen to be holding, instead of a habit you can carry with you.

How it works

1. Pick the surface that matches the device. Desktop apps for macOS, Windows, and Linux. Mobile apps on the App Store and Google Play. Apple Watch app pre-installed with the iPhone build. Chrome extension from the Chrome Web Store. Obsidian plugin from Community Plugins. REST API on the Pro tier.
2. Sign in once with the same account. Sign in with Apple, Google, GitHub, or email magic-link - whichever you already use. Pro covers up to 10 devices, with no separate "work seat" vs "personal seat" - your Windows work laptop and your Mac at home count against the same cap.
3. On desktop, press ⌃+⌥+R on Mac or Ctrl+Alt+R on Windows / Linux. Speak. Press it again. Clean text inserts at your cursor in whatever app has focus.
4. On iPhone or Android, tap the mic (or the Android home-screen widget, which records directly via MediaRecorder - not a launcher). Speak. The note saves with auto-copy to clipboard. On Apple Watch, tap the wrist, talk, stop - the watch records directly to the cloud over its own connection (Wi-Fi, cellular, or paired iPhone tether). On a cellular Apple Watch, recording works fully standalone; transcripts appear on your iPhone via cloud sync.
5. In Chrome, the extension inserts into any web text field - Gmail, Slack web, ChatGPT, Claude, Notion. In Obsidian, the plugin drops cleaned text into the current note.
6. Cloud sync is off by default. Turn it on if you want notes to flow across devices; leave it off and each device works fully on its own. When it is on, the data is end-to-end encrypted with AES-256-GCM and an Argon2id passphrase you set yourself - only you can decrypt your notes.
7. On Pro, call the REST API from your own scripts or tooling. Same transcription pipeline, programmatic.

Cross-platform sync without giving up your data

Cloud sync is opt-in, not the default. If you turn it on, notes are encrypted on the device before they leave it. The key is derived from a passphrase you set, with Argon2id (memory 65536 KiB, iterations 3, parallelism 4), and the encrypted blobs travel over modern TLS - TLS 1.3 on macOS and iOS, certificate-pinned on iOS and Android, and TLS 1.3 where the mobile OS supports it. We cannot read what is in your account on our servers. If you leave sync off, every device still works fully - transcription, cleanup, custom vocabulary, the whole pipeline - just without the cross-device mirror.

Voice input itself works in 99 languages on every surface. The mobile UI (iPhone, iPad, Apple Watch, Android) is localized in 28 languages. The desktop apps, Chrome extension, and Obsidian plugin present their UI in English only - but a Russian-speaking engineer on Linux still gets Russian transcription, with the menus in English.

On desktop, every diagnostic surface is user-initiated: the OS-level "share with developer" prompt that macOS shows after a crash (which you have to accept), or a manual "Send Diagnostic Report" button you click yourself. On mobile, we use Firebase Analytics for sign-in and subscription events only - we do not track your transcription content, recordings, or sharing behavior, and on Android we explicitly omit the AD_ID permission. Your audio and text never enter our analytics pipeline.

Audio retention is the same everywhere: streamed to Groq, processed, dropped immediately after processing, within 1 second. No persistent audio copy. Same number on macOS, Windows, Linux, iOS, Android, Chrome, and the API.

Linux without the asterisk

Most voice apps in this category say "we support Linux" and ship a single AppImage. AICHE ships four package formats so it actually lands cleanly on the distros engineering teams run:

• .deb for Debian, Ubuntu, Linux Mint, elementary OS, Pop!_OS
• .rpm for Fedora, RHEL, Rocky Linux, AlmaLinux
• AppImage for any glibc-based distro without an install step
• Flatpak for sandboxed, distro-agnostic install via Flathub

Each format has its own runtime expectations and its own test surface. System libraries on .deb and .rpm, sandbox interfaces on Flatpak, bundled libraries on AppImage. The hotkey (Ctrl+Alt+R) registers system-wide on X11 without additional setup. On Wayland-based desktops (GNOME 41+), you may need to grant global hotkey permission once. After that, the workflow on Linux is the same as on Mac and Windows: position cursor, hotkey, speak, hotkey, text appears.

This matters for one specific reason: most companies don't run one operating system. Office staff on Windows. Executives on Mac. The infra and ops teams on Linux. Field staff on iOS and Android. AICHE covers the whole zoo with one product, one subscription model, and one cleanup quality. The team that builds the infrastructure gets the same voice layer as the people writing the email.

For developers: the REST API and the dev profile

Pro tier exposes the transcription pipeline as a REST API you can call from your own scripts. Self-serve, not enterprise-gated. The same Whisper-plus-polish pipeline that the desktop and mobile apps use - hallucination filter, filler removal, custom vocabulary enforcement, Groq-hosted LLM polish - runs behind the API. ~3 seconds for 15 minutes of audio.

Pair the API with the Software Development profile (Pro), which tunes recognition for code identifiers, kebab-case CLI flags, snake_case variables, library names, and the kind of speech you actually use when dictating about software. Your 50-entry custom vocabulary syncs across every surface, so a service name you teach AICHE on the Mac is spelled correctly on the Linux box and via the API.

One subscription, both halves of your life

The device cap is per account, not per OS - up to 10 devices on Pro. A typical Pro setup might look like Mac at home, Windows at work, Linux dev box, iPhone, iPad, Apple Watch, Android tablet, Chrome on a second laptop, Obsidian on the same Mac, and a script calling the API. Same login, same custom vocabulary, same cleanup quality. If you want notes to follow you, turn sync on. If you keep work and personal separate, leave sync off and the device cap still buys you the same voice layer on every machine.

Bought it on iPhone, use it on Mac, on Linux, in Chrome - one subscription follows you everywhere. The Manage Apps screen in your account dashboard shows every device signed into your subscription and lets you revoke any session from one place, so a retired laptop or a borrowed phone never holds a seat hostage.

Free trial is 7 days, no credit card. Personal annual works out to $3.99/mo. Pro annual is $8.33/mo.

Tips

Set up the high-traffic devices first. Install on the two or three machines you actually dictate from daily. Add the others as you encounter them. There is no points-for-completeness benefit to installing on all ten the first weekend.

Use the Chrome extension as the bridge for web-only work. If your day is Gmail, Slack, Linear, Notion, and an AI chat tab, the Chrome extension covers all of that on a machine where you may not want the full desktop app.

Custom vocabulary, once. Add your 50 entries on any device. They sync to every other device on your account, including the API. Don't re-enter them on each platform.

Leave cloud sync off if you don't need it. It costs nothing to keep notes local-only and each device still works fully. Turn it on when you actually need cross-device access, not by default.

Apple Watch is standalone on cellular. Watch records directly to the cloud over its own connection - Wi-Fi, cellular, or paired iPhone tether. On a cellular Apple Watch (Series 5+ LTE), leave your iPhone at home and the recording uploads from your wrist. Transcripts appear on your iPhone via cloud sync.

Works With Everything

AICHE's global hotkey inserts text into whatever has focus. Some of the 100+ supported tools:

• AI agents: Claude Code, Cursor, ChatGPT, Gemini, GitHub Copilot
• IDEs: VS Code, JetBrains, Xcode, Android Studio
• Messaging: Slack, Discord, Microsoft Teams
• Notes: Notion, Obsidian, Google Docs
• Terminals: Terminal, iTerm2, Warp

See all integrations

Result: the voice layer stops being a function of which device you happen to be holding. One subscription, one login, one cleanup pipeline across 9 platforms, with API access, audio dropped immediately after processing, within 1 second and sync you can leave off.

Do this now: install AICHE on the next device you pick up that isn't the one you usually dictate from, sign in with the same account, and confirm your custom vocabulary is already there.