macSTT

macOS-native speech-to-text app for dictation on macOS.

The app listens for a configured trigger, captures microphone audio, runs local speech-to-text, and types the final transcript at the current cursor location.

Download the latest macSTT release

Requirements

• macOS 15+
• Xcode 16+
• XcodeGen: brew install xcodegen
• Fastlane: brew install fastlane

Development

# Build and run (kills any running instance first)
./run.sh

# Build only
swift build

# Run tests
swift test

Development builds run via swift run (bare binary, not .app bundle).

run.sh will optionally codesign the debug binary if CODE_SIGN_IDENTITY is set:

CODE_SIGN_IDENTITY="Apple Development: Your Name (TEAMID)" ./run.sh

Features like Launch at Login require a bundled .app; in dev mode the control is shown but disabled.

Runtime Notes

• Model status is shown in the settings window as Initializing, Checking Models, Downloading Models, Loading Models, Ready, or Error.
• Models are stored under ~/Library/Application Support/macSTT/Models.
• The app needs microphone permission for capture.
• The app needs Accessibility permission to type transcriptions into other apps.

Build .app

# Generate Xcode project + build signed .app
fastlane mac build

Output: build/macSTT.app. Open it directly or drag to /Applications/.

This path requires a local DEVELOPMENT_TEAM and a matching Developer ID signing identity. Sparkle is disabled for this local build path by default.

Release

# Bump patch version, build, notarize, create DMG, publish to GitHub
fastlane mac release

This bumps the version in project.yml, builds a signed .app, notarizes it with Apple, creates a DMG, publishes to the itsklimov/macSTT GitHub releases, and updates the Sparkle appcast on the repository's gh-pages branch.

Official release credentials are only needed for this path: DEVELOPMENT_TEAM, FASTLANE_USER, APPLE_ID, FASTLANE_APPLE_APPLICATION_SPECIFIC_PASSWORD, SPARKLE_PUBLIC_ED_KEY.