claude-code-tts-hooks

TTS Server

Audio feedback for code agents using text-to-speech. Hear summaries of responses and permission requests.

Supports Claude Code (via shell hooks) and OpenCode (via TypeScript plugin).

Features

- Response summaries - After the agent responds, hear a TTS summary of what it did
- Permission announcements - Hear what permission the agent is requesting before you approve
- Interrupt support - New audio cancels currently playing audio with a transition chime
- SSH support - Works on remote servers via reverse tunnel
- REST API - Control the TTS server programmatically

Audio Examples

Response summary - After Claude completes a task

https://github.com/user-attachments/assets/217b3273-092b-4a70-9b6a-1ed45777b60a

Permission request - Before a tool requires approval

https://github.com/user-attachments/assets/df42f17a-2c52-4346-91df-0510c49a8655

Requirements

- uv (Python package manager)
- Groq API key (free tier works fine) OR Ollama for local inference
- A supported code agent: Claude Code or OpenCode

Installation

1. Install Dependencies

2. Configure the Server

Copy the example environment file and add your API key:

Edit .env and set your Groq API key:

See Configuration below for all available options.

3. Start the TTS Server

4. Set Up Your Code Agent

Choose your agent below:

Claude Code

Install the Hooks

Configure Claude Code Hooks

Add to ~/.claude/settings.local.json:

Test It

You may need to load claude and check the hooks are loaded with the /hooks command.

OpenCode

Install the Plugin

No extra dependencies needed — the plugin uses only the built-in @opencode-ai/plugin types.

Test It

Start the TTS server, then launch OpenCode in your project. Send a prompt and you should hear audio when the agent finishes responding.

Architecture

API…