hey-claude

Hey Claude

Your AI agent talks to you.

A Claude Code plugin that gives Claude a voice. Instead of staring at your terminal waiting, Claude literally speaks to you — tells you what it did, what it needs, and when it's your turn.

Why?

You're working on something else. Claude finishes a complex refactor. Instead of a silent cursor blink, you hear:

"Auth refactor done. Changed 3 files. Your turn."

You're grabbing coffee. Claude needs permission:

"Hey, need your go-ahead to run a shell command."

You're in another window. Claude hit an error:

"Hit a type error in dashboard. Need your eyes on this."

Zero cost. Works with built-in system TTS or Microsoft's neural voices via edge-tts (free, no API key).

How It Works

1. When a session starts, Claude gets a tiny instruction to include <voice> tags in its responses
2. Claude naturally writes <voice>short contextual message</voice> as it works
3. When Claude stops, a hook extracts the voice tag and speaks it
4. For system notifications (permissions, dialogs), fallback messages play automatically

Claude finishes work
    → Stop hook fires
    → Reads transcript, finds: <voice>Dashboard done. Tests passing.</voice>
    → Speaks: "Dashboard done. Tests passing."
    → You hear it from across the room

Install

Add "hey-claude@aryanxpatel": true to your enabledPlugins in ~/.claude/settings.json.

Then run /voice setup in Claude Code to pick your personality, TTS engine, and voice.

TTS Engines

Engine	Quality	Latency	Offline	Setup
Built-in (default)	Basic	Instant	Yes	Zero — uses Windows SAPI, macOS say, Linux espeak
Edge TTS (recommended)	Neural, human-like	~1 second	No (needs internet)	pip install edge-tts

Edge TTS Voices

Edge TTS gives you access to Microsoft's neural voices — the same ones used by Edge browser's "Read Aloud". 300+ voices in 70+ languages, completely free, no API key.

Best English male voices:

Voice	Vibe
en-US-ChristopherNeural	Clear, authoritative, JARVIS-like
en-US-AndrewNeural	Smooth, professional presenter
en-US-GuyNeural	Warm, natural, conversational
en-US-BrianNeural	Casual, friendly, younger
en-US-RogerNeural	Deep, commanding

To install: pip install edge-tts

Personalities

Casual Buddy (default)

"Hey, auth refactor done. 3 files changed." "Need your go-ahead on this." "Done! What's next?"

JARVIS

"Sir, refactoring complete. 3 files modified." "Authorization required, sir." "All tasks completed. Awaiting instructions, sir."

Professional

"Auth refactor complete. 3 files changed." "Permission needed to proceed." "Task complete. Ready for review."

Custom

Create your own personality with custom messages. See Configuration.

Commands

Command	Description
/voice	Show current settings
/voice setup	Interactive setup wizard (personality, engine, voice, speed, volume)
/voice test	Test current voice
/voice mute	Mute (DND mode)
/voice unmute	Unmute
/voice personality jarvis	Switch personality
/voice volume 80	Set volume (0-100)
/voice speed 3	Set speech rate (-5 to 5)
/voice voices	List available voices
/voice voice <name>	Set specific voice

Cross-Platform

Platform	Built-in Engine	Edge TTS
Windows	System.Speech (SAPI)	Yes
macOS	say command	Yes
Linux	espeak / piper / festival	Yes (needs mpv or ffplay for MP3)

Linux Setup

# Built-in TTS
sudo apt install espeak

# Edge TTS (recommended)
pip install edge-tts
sudo apt install mpv    # for MP3 playback

Configuration

Settings are stored in config/voice-config.json:

{
  "personality": "casual",
  "engine": "builtin",
  "volume": 100,
  "rate": 2,
  "muted": false,
  "voice": null,
  "edge_voice": "en-US-ChristopherNeural",
  "edge_rate": "+10%"
}

Field	Description	Values
personality	Voice personality preset	casual, jarvis, professional, or custom
engine	TTS engine to use	builtin or edge-tts
volume	Playback volume	0-100
rate	Speech rate (built-in only)	-5 to 5
voice	Specific built-in voice name	System voice name or null for default
edge_voice	Edge TTS voice ID	e.g., en-US-ChristopherNeural
edge_rate	Edge TTS speed adjustment	e.g., +10%, -20%, +0%

Custom Personality

Add your own personality to config/personalities.json: