EchoType

English | 中文

EchoType is a native offline speech-to-text menubar app for macOS.

It lets you start recording with a global hotkey, transcribes speech into text locally, copies the result to the clipboard, and can automatically paste it into the current input field when the required permissions are granted.

Features

• Start or stop recording with a global hotkey in one step (default: ⌥ ⌘ Space, configurable directly in Settings)
• Fully local offline transcription with WhisperKit and Qwen3-ASR (speech-swift / MLX)
• Two text injection modes: clipboard only, or clipboard plus auto-paste with automatic fallback
• Settings page includes hotkey editing, permission guidance, model install/remove, and language hints

Auto-paste note: when Clipboard + Auto Paste is enabled, EchoType pastes into whichever app is frontmost at transcription completion time. The transcribed text is also placed on the macOS clipboard before the paste shortcut is sent.

System Requirements

• macOS 14+
• Apple Silicon (M1 or newer)
• Microphone permission is required for recording
• Accessibility permission is optional for auto-paste; clipboard copy still works without it

Installation

Option A: Homebrew (Recommended)

brew tap smoosex/tap
brew install --cask echotype

Or install in one line:

brew install --cask smoosex/tap/echotype

Option B: GitHub Releases (DMG)

1. Download EchoType-<version>.dmg
2. Open it and drag EchoType.app into Applications
3. Launch the app and grant the requested permissions on first run

Note: the public build currently ships with ad-hoc signed executables and resource bundles. Developer ID signing and notarization are not enabled yet. After downloading or updating the app, run the following command, then re-grant microphone and accessibility permissions. For accessibility permission, you may need to remove the old entry first and then authorize it again.

sudo xattr -dr com.apple.quarantine "/Applications/EchoType.app"

First Launch

1. On first launch, EchoType opens the Welcome Guide automatically so you can complete microphone/accessibility permission setup and read the usage notes
2. You can enable Don't show this guide again so it will not appear on startup in the future. You can still reopen it from the Welcome Guide menu
3. Click Start Using EchoType to open the Settings page automatically
4. In the General tab, click the hotkey recorder field and press your preferred shortcut directly. If registration fails, EchoType shows the reason and falls back to the default ⌥⌘Space
5. In the Models tab, choose and install a model
6. Return to the main app, press the hotkey to start recording, then press it again to stop and wait for transcription

Models

• WhisperKit: for Whisper models
• speech-swift: for Qwen3-ASR models (Apple Silicon GPU / MLX)

Engine projects used in EchoType:

• Whisper models are powered by WhisperKit
• Qwen3-ASR models are powered by speech-swift

Currently available models:

• Whisper Tiny
• Whisper Base
• Whisper Large v3
• Qwen3-ASR 0.6B
• Qwen3-ASR 1.7B

After installation, EchoType automatically switches to the corresponding engine based on the selected model.

Local Development

swift build
scripts/build_mlx_metallib.sh debug
swift run

Note: speech-swift depends on mlx.metallib. After the first local development build, run scripts/build_mlx_metallib.sh debug once. If your system currently uses only Command Line Tools, install the full Xcode app first, then run xcodebuild -downloadComponent MetalToolchain.

Complete Uninstall

scripts/uninstall.sh

Common options:

scripts/uninstall.sh --dry-run
scripts/uninstall.sh --yes

Privacy

• Audio and transcription text are not uploaded to external services by default
• Logs are used for diagnostics and do not record raw audio content