VoiceScribe

Let your voice do the work.

Features

• Global hotkey recording - Press Option-Shift-Space to record from anywhere (customizable)
• Dual transcription engines - Choose between privacy-focused local WhisperKit (on-device, Apple Silicon only) or cloud-based OpenAI Transcription
• Multiple AI models - Download and switch between WhisperKit models (Base/Small/Medium) or select OpenAI models (Whisper V2/GPT-4o/GPT-4o Mini)
• AI-powered enhancement - Optional post-processing to add perfect punctuation, capitalization, and formatting:
  • Local mode: 100% private on-device enhancement using MLX models (Qwen 2.5/Llama 3.2/Phi-3.5)
  • Cloud mode: Fast enhancement using OpenAI GPT-4o-mini
• Auto-paste - Transcriptions paste directly into your active app
• Transcription history - Review past transcriptions anytime
• Secure storage - API keys encrypted in macOS Keychain

Screenshots

Recording Workflow

History

Settings

Menu Bar

Installation

Homebrew (Recommended)

brew install eddmann/tap/voicescribe

Manual Download

1. Download the latest release from GitHub Releases
2. Unzip and move VoiceScribe.app to Applications
3. Double-click to open

The app is signed and notarized by Apple, so it will open without any security warnings.

Usage

• Press Option-Shift-Space to open the recording window
• Press Space to start/stop recording
• Transcription automatically copies to clipboard and pastes if enabled

First Launch

1. VoiceScribe appears in your menu bar as a waveform icon
2. Click and select "Settings" to choose your transcription service:
   • Local WhisperKit (recommended) - Download a model to get started
   • OpenAI Transcription - Add your API key from OpenAI and select a model (Whisper V2, GPT-4o, or GPT-4o Mini)
3. Optionally enable "Enhance transcriptions" for AI-powered post-processing:
   • WhisperKit users can choose an MLX model for 100% private local enhancement
   • OpenAI users get cloud-based enhancement via GPT-4o-mini
4. Enable "Auto-paste" in Preferences and grant Accessibility permission if desired
5. Press Option-Shift-Space to start your first recording

Requirements

• macOS 14.0 (Sonoma) or later
• Apple Silicon (M-series) Mac for local WhisperKit transcription and MLX enhancement
• Intel Macs supported with OpenAI Transcription API

Privacy

VoiceScribe gives you full control over your data:

Local Mode (WhisperKit + MLX)

• 100% private - All transcription and AI enhancement happens on-device
• No network transmission - Your audio and text never leave your Mac
• Ideal for sensitive content - Perfect for lawyers, doctors, journalists, and anyone handling confidential information
• Free - No API costs

Cloud Mode (OpenAI)

• Audio transmission - Audio files sent to OpenAI servers for transcription
• Optional text enhancement - If post-processing enabled, transcribed text sent to OpenAI for formatting
• API costs - Approximately $0.006-0.024/minute for transcription, ~$0.004 additional if enhancement enabled
• Review OpenAI's Privacy Policy for details

Security Features

• API keys encrypted in macOS Keychain with device-only access
• Audio files stored in temporary directory and automatically deleted after transcription
• Transcription history stored locally with SwiftData
• No telemetry or analytics