Dictator is a lightweight macOS menubar application that provides near-instant speech-to-text transcription using OpenAI's Whisper API. Powered by Hammerspoon, it offers a seamless workflow for dictating text directly into any application.
Dictator converts speech to text with professional accuracy using the latest Whisper AI models. It sits quietly in your menubar, ready to transcribe your thoughts the moment you press a key.
- Fast & Reliable: Near-instant transcription via Groq or OpenAI.
- BYOK (OpenAI Standard): Support for any OpenAI-compatible API provider.
- Privacy First: Audio is processed only by your chosen API provider.
- Smart Formatting: Optional AI-powered grammar and punctuation refinement.
- Native Experience: A sleek macOS menubar app that works with any text field.
- Context Awareness: (New) Automatically detects the active app and window to adapt the correction style accordingly.
To get started, ensure you have Hammerspoon and SoX installed:
brew install --cask hammerspoon sox
# 1. Clone the repository
git clone https://github.com/Glossardi/Dictator-Speech-to-Text.git ~/Documents/Dictator
# 2. Run the installer
cd ~/Documents/Dictator
./install.shThe installer will guide you through the initial setup, including API key configuration and system permissions.
Stay current with the latest transcription models and features:
Click the Dictator icon (ποΈ) and select Update Dictator.... This handles everything automatically.
cd ~/Documents/Dictator
make update- Grant Permissions: Ensure Hammerspoon has Accessibility and Microphone access in System Settings > Privacy & Security.
- Setup API: Click the ποΈ icon > Settings > Set API Key.
- Dictate: Hold the
Fnkey, speak, and release to transcribe directly into your focused app.
If the automated script fails or you prefer full control, use these commands to manually install or update:
# 1. Install dependencies
brew install --cask hammerspoon sox
# 2. Copy application files
mkdir -p ~/.hammerspoon
cp -v ~/Documents/Dictator/*.lua ~/.hammerspoon/
# 3. Reload Hammerspoon
# Click icon (ποΈ or π¨) > Reload Config- ποΈ Hold-to-Record: Minimalist workflow with customizable hotkeys.
- β‘ Performance: Optimized for Groq (transcription often under 500ms).
- π BYOK: Compatible with all OpenAI-standard API providers.
- π― Precision: Support for custom technical glossaries to avoid mistakes.
- π§ Context Awareness: Detects active apps (Slack, Mail, VS Code), window titles, and integrates your personal glossary to provide smarter, context-aware corrections that respect your specific terminology.
- π Global: Multi-language support via Whisper-v3 Large models.
This project is licensed under the MIT License.