Skip to content

User Guide

Everything you need to know to get started with Airgap Voice.

Getting Started

Airgap Voice is a menu bar application that transcribes speech directly at your cursor — in any application. All processing runs locally on your Mac’s Apple Silicon GPU. No audio ever leaves your machine.

Requirements

  • macOS 15 (Sequoia) or later
  • Apple Silicon Mac (M1, M2, M3, M4 or later)
  • Microphone access (built-in or external)
  • Accessibility permission (for cursor text insertion)

First Launch

  1. Open Airgap Voice from your Applications folder
  2. Grant microphone and accessibility permissions when prompted
  3. The app icon appears in your menu bar
  4. Click the icon or use the keyboard shortcut to start recording

Live transcription directly into Microsoft Word — text appears at the cursor position in real time.

Click the Airgap Voice icon in the menu bar to open the main menu. From here you can start recording, change language, select a microphone, open settings, access support, or quit the application.

The main menu provides quick access to all key functions.

Recording

To start recording, click Start Recording in the menu or use the keyboard shortcut. The menu bar icon turns green to indicate active recording.

Speak naturally — Airgap Voice transcribes your speech in real time and inserts the text at your cursor position in whatever application is focused. When you’re done, click Stop Recording or use the shortcut again.

Tip: Place your cursor in the target application before starting a recording. The text will appear exactly where your cursor is.

Accessibility

Airgap Voice is designed for users with motor impairments, RSI, repetitive strain injuries, or typing difficulties. Dictate text entirely offline and insert it at your cursor in any application.

VoiceOver & System Accessibility

Airgap Voice fully supports VoiceOver. All controls are labeled and state changes are announced. To enable VoiceOver or other accessibility features on your Mac, open System Settings > Accessibility.

VoiceOver Announcements

  • Announcement Detail — Controls how many events are announced by VoiceOver. Essential announces only recording start/stop and errors. Detailed announces all events.
  • Announce Inserted Text — When enabled, VoiceOver reads back a preview of the transcribed text after it is inserted.

Accessibility settings with VoiceOver announcement controls.

Custom Dictionary

Add domain-specific terms to improve recognition accuracy. Custom vocabulary terms are used for vocabulary boosting during transcription, helping the model correctly recognize specialized terminology.

Toggle Enable Custom Vocabulary to activate the dictionary. Add terms using the text field at the bottom. Remove terms with the delete button next to each entry.

Add words or phrases that are frequently misrecognized by the transcription model.

Note: Larger dictionaries may add a slight delay to the transcription process.

Language

Configure spoken language detection, preferred languages, and filler word removal.

Auto-Detect

When enabled, Airgap Voice automatically detects the spoken language. No manual selection needed. For Asian languages such as Malay or Indonesian, disable auto-detect and select the language manually to ensure correct recognition.

Preferred Languages

Optionally select preferred languages to improve detection accuracy. This helps the model when multiple languages could match.

Filler Words

Enable filler word removal to automatically strip words like “um”, “uh”, “like”, “hmm” from your transcriptions. You can customize the list of filler words to remove.

Language settings with auto-detection, preferred languages, and filler word removal.

Model

Choose between two transcription models and two transcription modes.

Turbo Model

Fastest transcription. Best for English and European languages. Works on any Mac.

Precision Model

Highest accuracy. Best for Asian languages and complex multilingual transcription.

Supported Languages

30 languages across both models. Turbo: Bulgarian, Croatian, Czech, Danish, Dutch, English, Finnish, French, German, Italian, Polish, Portuguese, Romanian, Russian, Slovak, Slovenian, Spanish, Swedish, Ukrainian. Precision: all Turbo languages except Bulgarian, Croatian, Slovak, Slovenian, Ukrainian — plus Cantonese, Chinese, Filipino, Hindi, Indonesian, Japanese, Korean, Malay, Thai, Turkish, and Vietnamese.

Transcription Mode

Streaming transcribes in real time as you speak. Text appears while you talk. Best for English. Batch records the full audio first, then transcribes after you stop. More accurate for non-English languages and longer dictation.

Model selection with supported languages and transcription mode.

Recording Settings

Configure microphone selection, auto-stop behavior, and end-of-speech detection.

Microphone

Select which microphone to use for recording. Supports built-in and external microphones.

Auto-Stop Recording

When enabled, recording automatically stops after a configurable silence duration. Choose from 5s, 15s, 30s, 45s, or 1 minute.

End-of-Speech Detection

Controls the silence duration required before finalizing speech. Lower values (0.2s) give snappier responses; higher values (1.0s) avoid mid-sentence splits. Default is 0.5s.

Recording configuration with microphone selection, auto-stop, and end-of-speech detection.

Shortcuts & Behavior

Configure keyboard shortcuts and application behavior.

Toggle Recording

Set a global keyboard shortcut to start and stop recording from any application. Default is ⌥R (Option + R).

Behavior

  • Confirm before quitting — Shows a confirmation dialog before quitting the app.
  • Launch at Login — Automatically starts Airgap Voice when you log in.

Keyboard shortcuts and application behavior settings.

Smart Formatting

Smart Formatting automatically converts spoken expressions to written form. Enable or disable individual formatting categories to match your workflow.

Numbers

Converts spoken numbers, ordinals, decimals, and fractions to their written form (e.g., “twenty three” → “23”).

Smart Formatting Numbers category with individual toggles for numbers, ordinal numbers, decimals, and fractions.

Date & Time

Formats spoken dates and times (e.g., “March fifteenth twenty twenty six” → “March 15, 2026”).

Smart Formatting Date & Time category with toggles for dates and times.

Money & Measures

Formats currency amounts and measurement units (e.g., “fifty dollars” → “$50”).

Smart Formatting Money & Measures category with toggles for currency and measurements.

Communication

Formats phone numbers, email addresses, and URLs spoken naturally.

Smart Formatting Communication category with toggles for phone numbers and emails/URLs.

Text

Handles punctuation, abbreviations, word substitutions, and address formatting.

Smart Formatting Text category with toggles for punctuation, abbreviations, word substitutions, and addresses.

Access support resources, report bugs, request features, and view legal documents — all directly from the app’s Settings panel.

  • User Guide — Opens this guide in your browser.
  • Email Support — Opens your email client with a pre-filled support address.
  • Report a Bug — Opens the bug report form on the website.
  • Request Feature — Opens the feature request form on the website.
  • Privacy Policy and Terms of Service — Opens the respective legal pages.

The Support panel provides direct access to help, bug reports, feature requests, and legal documents.

About

View the current app version, build number, license status, and check for updates.

  • Version — Shows the installed version and build identifier.
  • License Status — Shows whether the app is activated, in trial mode, or unlicensed.
  • Check for Updates — Opens the update check page in your browser (the app itself makes no network connections).

About panel showing version, license status, and update check link.

Designed for mission critical security