Don't just type to your AI coding tool. Talk to it.
What is voCLI?
You speak. Claude listens. Claude responds. You hear it.
voCLI adds a voice layer to Claude Code. Everything runs locally on your machine — no audio is sent to the cloud, ever.
You speak --> Mic --> faster-whisper (STT) --> Text to Claude
Claude responds --> Kokoro TTS --> Audio plays through speakers
Highlights
- Privacy-first — all audio processing stays on your machine
- Works offline — after initial model download, no internet needed
- Personalized — set your name and the assistant's name
- High-quality voice — natural-sounding Kokoro TTS with 27 voices to choose from
- Smart performance — auto-detects Apple Silicon vs Intel for optimal speed
- Remote-ready — use voice even when Claude Code runs on a remote VM
Quick Start
Recommended: Claude Code Plugin Marketplace
The easiest way to install voCLI:
# Step 1: Add the marketplace
/plugin marketplace add voCLI/voCLI
# Step 2: Install the plugin
/plugin install vocli@vocli
# Step 3: Run the setup wizard
/vocli:install
# Step 4: Start talking!
/vocli:talk
The install wizard walks you through everything: dependencies, models, and configuration.
Alternative: Manual Install via UV
# Install UV package manager (if needed)
curl -LsSf https://astral.sh/uv/install.sh | sh
# Add voCLI as an MCP server
claude mcp add --scope user vocli -- uvx --refresh vocli serve
# Restart Claude Code, then run:
/vocli:install
Local Install
For running Claude Code and voCLI on the same machine (Mac or Linux with mic/speakers).
The /vocli:install wizard handles everything:
| Step | What happens |
|---|---|
| 1 | Checks Python 3.10+ and ffmpeg |
| 2 | Installs faster-whisper and kokoro-onnx |
| 3 | Downloads Kokoro voice model (~325MB) and Whisper model (~500MB) |
| 4 | Detects your CPU architecture (Apple Silicon / Intel / ARM) |
| 5 | Lets you choose model size (tiny for speed, small for accuracy) |
| 6 | Asks for your name, assistant name, and preferences |
Performance note: On Apple Silicon, voCLI automatically uses float16 for faster speech recognition. On Intel, it uses int8.
Remote Install
For running Claude Code on a remote VM while keeping voice on your local machine.
Why?
Remote VMs have no microphone or speakers. voCLI's remote mode runs the audio servers on your local machine and connects them to Claude Code on the VM via the network.
How it works
Your Mac/Linux (local) Remote VM
+------------------+ +------------------+
| STT Server | <-- network | Claude Code |
| (port 2022) | -----------> | + voCLI Plugin |
| | | |
| TTS Server | <-- network | /vocli:talk |
| (port 8880) | -----------> | just works! |
| | | |
| Mic + Speakers | | No audio deps |
+------------------+ +------------------+
Setup
Install the plugin on your remote VM, then run the remote setup wizard:
The /vocli:remote-install wizard handles everything:
From starting the servers on your local machine, entering the URLs, and configuring your preferences.
Tip: If your VM can't reach your local machine directly, use SSH port forwarding:
ssh -R 2022:localhost:2022 -R 8880:localhost:8880 your-vmThen use
http://localhost:2022andhttp://localhost:8880as the URLs.
Commands
| Command | Description |
|---|---|
/vocli:install |
Install dependencies, download models, configure (local mode) |
/vocli:remote-install |
Set up remote mode with external STT/TTS servers |
/vocli:config |
Change assistant name, your name, preferences |
/vocli:talk |
Start a voice conversation |
MCP Tools
voCLI runs as an MCP server with three tools:
| Tool | What it does |
|---|---|
talk |
Speak a message and optionally listen for a reply. Auto-starts servers if needed. |
status |
Check health of STT/TTS servers and show current config. |
service |
Start, stop, or restart STT/TTS servers. |
Requirements
| Requirement | Details |
|---|---|
| Python | 3.10 or higher |
| OS | macOS (Apple Silicon or Intel) or Linux |
| Disk space | ~900MB for models |
| Audio | Microphone + speakers |
License
MIT
Metadata
Release files for vocli 0.2.5
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| vocli-0.2.5.tar.gz | 132.7 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| vocli-0.2.5-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 155.7 kB
Release files / vocli-0.2.5.tar.gz
| Download URL | vocli-0.2.5.tar.gz |
|---|---|
| Size | 132.7 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
8091565b1c1ce55bf10bd9b73e19ddb8689cdb160cceb3f5ba9dab267479f0a8
|
|
BLAKE2b-256 checksum How to use checksums |
992d6248aa00a5ec7cc916fc6d8a6d98f5adc471ad870e000c555975248f954a
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Apr 13, 2026.
Transparency logRelease files / vocli-0.2.5-py3-none-any.whl
| Download URL | vocli-0.2.5-py3-none-any.whl |
|---|---|
| Size | 23.0 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
88fd90a198e92ae66de654e3306087b16c3180c438fd6979696928603faed797
|
|
BLAKE2b-256 checksum How to use checksums |
3686210f9fb4d9fba694632789c4d61a6ab2b5decb4a102d9878c8d01a3c65b1
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Apr 13, 2026.
Transparency log