Developers: MCP server, Agent Skills, and CLI

Wire Voice Studio into Claude Code, Codex, Cursor, or any agent. Text to speech, voice cloning, voice design, and transcription as native tools, running locally on your Mac.

Before you start

  • Voice Studio for Mac installed and signed in (macOS 12+ on Apple Silicon).
  • An active subscription or free trial. Everything runs on-device; no audio leaves your Mac.
  • Python 3.11+ (for pip) or uv installed (recommended, gives you uvx and pipx-style runs).

The integrations talk to the app's local sidecar and launch Voice Studio automatically if it isn't already running.

MCP server

The voicestudio-mcp package exposes Voice Studio as Model Context Protocol tools, so any MCP-aware agent can generate speech, clone and design voices, and transcribe audio. No skill or plugin required: once connected, the agent gets Voice Studio's built-in usage guide and knows how to drive each tool.

One command in your terminal:

claude mcp add voicestudio -- uvx voicestudio-mcp

Restart Claude Code. Ask it to “list Voice Studio voices” to confirm the connection.

Prefer a global install? pip install voicestudio-mcp then use voicestudio-mcp as the command.

Tools

vs_generate_speech

Text to speech with a built-in voice

vs_generate_with_profile

Speak with a saved cloned voice

vs_clone_voice

Clone a voice from a reference sample

vs_design_voice

Design a voice from a description

vs_transcribe

Transcribe an audio file

vs_list_voices

List built-in voices and saved profiles

vs_list_history

Browse recent generations

vs_download_from_history

Re-download a past generation

vs_account_status

Plan, tier, and free-generation balance

vs_current_service

Which model is loaded right now

vs_free_models

Unload all models to free memory

vs_mcp_enabled

Check whether the MCP server is enabled

Agent Skills

Agent Skills are portable folders of know-how that any compatible agent can load (Claude Code, Codex, Cursor, and more, per the open Agent Skills standard). Voice Studio ships two: voicestudio-mcp (MCP setup and recipes) and voicestudio-cli (shell and batch recipes). The skill carries the workflow; the MCP server carries the actions, so agents without skills still work from the MCP server alone.

Install by dropping a skill folder into your agent's skills directory:

Claude Code (personal skills)
git clone https://github.com/aiworkmagic/voicestudio
cp -r voicestudio/.claude/skills/voicestudio-mcp ~/.claude/skills/
cp -r voicestudio/.claude/skills/voicestudio-cli ~/.claude/skills/

For a project, use .claude/skills/ instead of ~/.claude/skills/. Codex and other agents read their own skills directory; the same SKILL.md folder is portable across all of them.

Command line

The voicestudio-cli package gives you the voicestudio command for scripts, pipelines, and CI. Every command takes --json for machine-readable output, and audio commands print the absolute path they wrote.

Install
pipx install voicestudio-cli
Try it
voicestudio say "Hello from Voice Studio" --voice Storyteller

Commands

voicestudio say

Generate speech from text

voicestudio clone

Clone a voice from a sample

voicestudio design

Design a voice from a description

voicestudio transcribe

Transcribe an audio file

voicestudio voices

List available voices

voicestudio account

Show plan and usage

Run voicestudio --help for the full command list.

Don't have the app yet? Get Voice Studio. Questions? Contact support.