punt-tts 0.6.1

Text-to-speech CLI, MCP server, and Claude Code plugin (ElevenLabs, AWS Polly, OpenAI)

punt-tts

Voice for your AI coding assistant.

When Claude Code finishes a task, hits an error, or needs your approval --- you hear it. No need to watch the terminal. Keep working; your assistant will tell you what happened.

Platforms: macOS

Quick Start

curl -fsSL https://raw.githubusercontent.com/punt-labs/tts/2d8922f/install.sh | sh

Restart Claude Code, then:

/notify y     # hear when tasks complete or need input
/recap        # spoken summary of what just happened

Manual install (if you already have uv)

uv tool install punt-tts
tts install
tts doctor

Verify before running

curl -fsSL https://raw.githubusercontent.com/punt-labs/tts/2d8922f/install.sh -o install.sh
shasum -a 256 install.sh
cat install.sh
sh install.sh

Features

• Notification layer --- spoken summaries when tasks finish, chimes when Claude needs input
• Session vibe --- /vibe sets the mood for all speech. Auto-mode reads session signals (test results, lint, git ops) and adapts the voice. Manual mode lets you set it yourself. ElevenLabs expressive tags ([weary], [excited], [sighs]) color every utterance.
• Three providers --- ElevenLabs (natural voice + expressive tags), OpenAI (low latency), AWS Polly (cost-effective)
• Opt-in only --- no audio until you enable it, no surprises
• Voice or chime --- /speak n switches to audio tones, no TTS API calls
• Graceful absence --- if punt-tts isn't installed, Claude Code works exactly as before
• MCP-native --- runs as a Claude Code plugin with slash commands and hooks

What It Looks Like

Enable notifications

> /notify y

Notifications: enabled (voice)
You will hear spoken summaries when tasks complete and chimes when Claude needs input.

Get a recap

> /recap

Speaking: "I refactored the authentication module into three files, added
comprehensive tests for the token refresh flow, and fixed a race condition
in the session middleware. All 47 tests pass."

Set the vibe

> /vibe banging my head against the wall

Vibe: banging my head against the wall → [frustrated] [sighs] [manual]

Auto-mode (default) reads session signals and adapts automatically --- after a string of test failures the voice sounds [weary], after a successful release it sounds [excited].

Switch to chime-only

> /speak n

Speak: off (chime only)
Notifications will use audio tones instead of voice.

Commands

Command	Purpose
/notify y	Speak on task completion and permission prompts
/notify c	Continuous --- also speak milestone updates during long tasks
/notify n	Off
/speak y	Notifications are spoken (default when /notify is on)
/speak n	Notifications are a chime --- no words
/recap	Spoken summary of Claude's last response
/say "text"	Speak arbitrary text aloud
/vibe <mood>	Set session mood --- voice adapts to match
/vibe auto	Auto-detect mood from session signals (default)
/vibe off	Disable vibe --- neutral voice
/voice on | /voice off	Enable/disable voice mode

Providers

punt-tts auto-detects the best available provider.

Provider	API Key	Default Voice	Best For
ElevenLabs	ELEVENLABS_API_KEY	matilda	Natural voice, expressive tags via /vibe
OpenAI	OPENAI_API_KEY	nova	Fast notifications, low latency
AWS Polly	AWS credentials	joanna	Cost-effective, no API key needed

Auto-detection order: ElevenLabs > OpenAI > Polly.

CLI

punt-tts is also a standalone TTS tool, independent of Claude Code.

tts synthesize "Hello world"                  # Synthesize with default provider
tts synthesize "Hello" --provider elevenlabs  # Use specific provider
tts doctor                                     # Check setup
tts install                                    # Install Claude Code plugin (marketplace)
tts uninstall                                  # Remove plugin and clean up
tts install-desktop                            # Register MCP server with Claude Desktop
tts serve                                      # Start MCP server (stdio)

Environment Variables

Variable	Description	Default
TTS_PROVIDER	Force a specific provider	auto-detect
TTS_MODEL	Model override	provider default
TTS_OUTPUT_DIR	Output directory	~/tts-output

Roadmap

Shipped

• Notification layer: /notify, /speak, /recap, Stop + Notification hooks
• Multi-provider TTS engine: ElevenLabs, AWS Polly, OpenAI
• Claude Code plugin: marketplace install, MCP server, slash commands
• CLI: synthesize, batch, pair, pair-batch, doctor
• Ephemeral output mode (.tts/ in cwd)
• Two-channel display: ♪ panel summaries with voice/provider context
• Playful stop hook phrases: randomized vocalization-themed messages ("Speaking my thoughts...", "Saying my piece...")
• Natural notification phrasing: randomized phrases for permission and idle prompts
• Audio playback serialization via flock --- concurrent utterances queue instead of overlapping
• ElevenLabs streaming API for lower time-to-first-audio
• Dev/prod namespace isolation for plugin testing (claude --plugin-dir .)
• /vibe with auto, manual, and off modes --- ElevenLabs expressive tags color every utterance
• Auto-vibe signal accumulator: test pass/fail, lint, git ops feed mood detection
• set_config MCP tool for atomic config mutations (replaces file-tool pattern)

Coming Soon

Feature	What It Does
Per-session voices	Each Claude Code session gets its own voice from a pool --- no more five matildas talking at once. /voice to audition and pick.
Screen reader	macOS VoiceOver and say fallback when no API key is configured

Documentation

Design Log | Changelog

Development

uv sync --all-extras                           # Install dependencies
uv run ruff check src/ tests/                  # Lint
uv run ruff format --check src/ tests/         # Format
uv run mypy src/ tests/                        # Type check (mypy)
uv run pyright src/ tests/                     # Type check (pyright)
uv run pytest tests/ -v                        # Test

License

MIT