Summary

Ghost Pepper offers a 100% local, hold-to-talk speech-to-text solution for macOS, designed for users prioritizing privacy and offline functionality. It allows users to transcribe spoken words and paste them into any application by simply holding the Control key, with no data ever leaving the machine.

How It Works

The application leverages open-source models running entirely on the user's Mac, utilizing Apple Silicon for processing. Speech transcription is handled by WhisperKit, while local LLM.swift models perform smart cleanup, removing filler words and self-corrections. Models are downloaded automatically and cached locally, ensuring offline capability and minimal setup friction.

Quick Start & Requirements

• Install: Download the GhostPepper.dmg file and drag the application to the Applications folder.
• Prerequisites: macOS 14.0+ and Apple Silicon (M1 or newer) are required. Microphone and Accessibility permissions must be granted.
• Build from Source: Clone the repository and build using Xcode (GhostPepper.xcodeproj).
• Documentation: No specific external documentation links are provided beyond the README.

Highlighted Details

• Hold-to-Talk: Press and hold Control to record; release to transcribe and paste.
• Local Processing: All models run locally on Apple Silicon, ensuring data privacy.
• Smart Cleanup: Integrated local LLM automatically removes filler words and corrects self-references.
• Menu Bar App: Operates discreetly from the menu bar without a dock icon.
• Customization: Users can configure cleanup prompts, select microphones, and toggle features.
• Privacy: No transcriptions are logged to disk; debug logs are in-memory only.

Maintenance & Community

No specific details regarding maintainers, community channels (e.g., Discord, Slack), or project roadmaps are present in the provided README.

Licensing & Compatibility

• License: MIT License.
• Compatibility: The MIT license permits commercial use and linking with closed-source applications. For managed enterprise devices, Accessibility permissions require IT administration via MDM profiles (e.g., Jamf, Kandji) using a Privacy Preferences Policy Control (PPPC) payload with specific Bundle ID (com.github.matthartman.ghostpepper) and Team ID (BBVMGXR9AY).

Limitations & Caveats

The application requires Accessibility permissions, which may necessitate administrative privileges or specific IT configurations on managed macOS devices. It is optimized for Apple Silicon hardware and macOS 14.0+.