$git clone https://github.com/mbailey/voicemode> Natural voice conversations with Claude Code (and other MCP capable agents)
| 1 | # VoiceMode |
| 2 | |
| 3 | > Natural voice conversations with Claude Code (and other MCP capable agents) |
| 4 | |
| 5 | [](https://pepy.tech/project/voice-mode) |
| 6 | [](https://pepy.tech/project/voice-mode) |
| 7 | [](https://pepy.tech/project/voice-mode) |
| 8 | |
| 9 | VoiceMode enables natural voice conversations with Claude Code. Voice isn't about replacing typing - it's about being available when typing isn't. |
| 10 | |
| 11 | **Perfect for:** |
| 12 | |
| 13 | - Walking to your next meeting |
| 14 | - Cooking while debugging |
| 15 | - Giving your eyes a break after hours of screen time |
| 16 | - Holding a coffee (or a dog) |
| 17 | - Any moment when your hands or eyes are busy |
| 18 | |
| 19 | ## See It In Action |
| 20 | |
| 21 | [](https://www.youtube.com/watch?v=cYdwOD_-dQc) |
| 22 | |
| 23 | ## Quick Start |
| 24 | |
| 25 | **Requirements:** Computer with microphone and speakers |
| 26 | |
| 27 | ### Option 1: Claude Code Plugin (Recommended) |
| 28 | |
| 29 | The fastest way for Claude Code users to get started: |
| 30 | |
| 31 | ```bash |
| 32 | # Add the VoiceMode marketplace |
| 33 | claude plugin marketplace add mbailey/voicemode |
| 34 | |
| 35 | # Install VoiceMode plugin |
| 36 | claude plugin install voicemode@voicemode |
| 37 | |
| 38 | ## Install dependencies (CLI, Local Voice Services) |
| 39 | |
| 40 | /voicemode:install |
| 41 | |
| 42 | # Start talking! |
| 43 | /voicemode:converse |
| 44 | ``` |
| 45 | |
| 46 | ### Option 2: Python installer package |
| 47 | |
| 48 | Installs dependencies and the VoiceMode Python package. |
| 49 | |
| 50 | ```bash |
| 51 | # Install UV package manager (if needed) |
| 52 | curl -LsSf https://astral.sh/uv/install.sh | sh |
| 53 | |
| 54 | # Run the installer (sets up dependencies and local voice services) |
| 55 | uvx voice-mode-install |
| 56 | |
| 57 | # Add to Claude Code |
| 58 | claude mcp add --scope user voicemode -- uvx --refresh --from voice-mode voicemode-mcp-launcher |
| 59 | |
| 60 | # Optional: Add OpenAI API key as fallback for local services |
| 61 | export OPENAI_API_KEY=your-openai-key |
| 62 | |
| 63 | # Start a conversation |
| 64 | claude converse |
| 65 | ``` |
| 66 | |
| 67 | For manual setup, see the [Getting Started Guide](docs/tutorials/getting-started.md). |
| 68 | |
| 69 | ## Features |
| 70 | |
| 71 | - **Natural conversations** - speak naturally, hear responses immediately |
| 72 | - **Works offline** - optional local voice services (Whisper STT, Kokoro TTS) |
| 73 | - **Low latency** - fast enough to feel like a real conversation |
| 74 | - **Smart silence detection** - stops recording when you stop speaking |
| 75 | - **Privacy options** - run entirely locally or use cloud services |
| 76 | |
| 77 | ## Compatibility |
| 78 | |
| 79 | **Platforms:** Linux, macOS, Windows (native or WSL), NixOS |
| 80 | **Python:** 3.10-3.14 |
| 81 | |
| 82 | ## Configuration |
| 83 | |
| 84 | VoiceMode works out of the box. For customization: |
| 85 | |
| 86 | ```bash |
| 87 | # Set OpenAI API key (if using cloud services) |
| 88 | export OPENAI_API_KEY="your-key" |
| 89 | |
| 90 | # Or configure via file |
| 91 | voicemode config edit |
| 92 | ``` |
| 93 | |
| 94 | See the [Configuration Guide](docs/guides/configuration.md) for all options. |
| 95 | |
| 96 | ## Permissions Setup (Optional) |
| 97 | |
| 98 | To use VoiceMode without permission prompts, add to `~/.claude/settings.json`: |
| 99 | |
| 100 | ```json |
| 101 | { |
| 102 | "permissions": { |
| 103 | "allow": [ |
| 104 | "mcp__voicemode__converse", |
| 105 | "mcp__voicemode__service" |
| 106 | ] |
| 107 | } |
| 108 | } |
| 109 | ``` |
| 110 | |
| 111 | See the [Permissions Guide](docs/guides/permissions.md) for more options. |
| 112 | |
| 113 | ## Local Voice Services |
| 114 | |
| 115 | For privacy or offline use, install local speech services: |
| 116 | |
| 117 | - **[Whisper.cpp](docs/guides/whisper-setup.md)** - Local speech-to-text |
| 118 | - **[Kokoro](docs/guides/kokoro-setup.md)** - Local text-to-speech with multiple voices |
| 119 | |
| 120 | These provide the same API as OpenAI, so VoiceMode switches seamlessly between them. |
| 121 | |
| 122 | ## Installation Details |
| 123 | |
| 124 | <details> |
| 125 | <summary><strong>System Dependencies by Platform</strong></summary> |
| 126 | |
| 127 | #### Ubuntu/Debian |
| 128 | |
| 129 | ```bash |
| 130 | sudo apt update |
| 131 | sudo apt install -y ffmpeg gcc libasound2-dev libasound2-plugins libportaudio2 portaudio19-dev pulseaudio pulseaudio-utils python3-dev |
| 132 | ``` |
| 133 | |
| 134 | **WSL2 users**: The pulseaudio packages above are required for microphone access. |
| 135 | |
| 136 | #### Fedora/RHEL |
| 137 | |
| 138 | ```bash |
| 139 | sudo dnf install alsa-lib-devel ffmpeg gcc portaudio portaudio-devel python3-devel |
| 140 | ``` |
| 141 | |
| 142 | #### macOS |
| 143 | |
| 144 | ```bash |
| 145 | brew install ffmpeg node portaudio |
| 146 | ``` |
| 147 | |
| 148 | #### NixOS |
| 149 | |
| 150 | ```bash |
| 151 | # Use development shell |
| 152 | nix develop |