Local Text-to-Speech for Claude Code and Codex
Give voice to your AI assistants with Kokoro.
1What is Aloud?
Aloud is a tool that reads Claude Code and Codex AI assistant responses aloud on macOS. This Python-based solution uses the Kokoro library for local text-to-speech conversion. After installation, assistant responses can be automatically vocalized.
2How It Works
The system requires Homebrew and Python 3.11/3.12. After installation, it downloads Kokoro model files and stores them in local cache. All processing happens locally ensuring data privacy. Users can listen to short summaries automatically and full responses with shortcuts.
3Installation Process
Aloud installs via pipx and configures various system settings. During installation, it creates configuration files, starts launchd service, and adds Hammerspoon shortcuts. Users need to grant accessibility permissions in macOS System Preferences.
4Usage Scenarios
Aloud is activated in Claude Code and Codex sessions with 'aloud on' command. Shortcuts allow listening to full responses or stopping playback. Each session is tracked separately and can be controlled from command line.
5Technical Details
The system maintains configuration files, cache and logs under ~/Library. It uses Kokoro speech engine, espeak-ng for phonemization, and Hammerspoon for shortcut management. Developers can modify voice, speed and retention settings via config.json.
6Why It Matters
Aloud makes AI assistant interactions more accessible. Its local operation ensures data privacy while the open-source license offers flexibility to developers. Such tools have potential to make AI experiences more inclusive and versatile.