voxtype-install

# voxtype Install Guide

voxtype is a standalone PyPI package for fully on-device voice
dictation on macOS. It types transcribed speech into whatever app has
focus. No cloud, no accounts, no audio leaving the machine.

## Install

One command; it picks the best engine for the machine automatically
(Apple-GPU `parakeet-mlx` on Apple Silicon, CPU `parakeet` elsewhere):

```bash
uv tool install voxtype
```

If the user doesn't have `uv`, either install it first
(`curl -LsSf https://astral.sh/uv/install.sh | sh`) or use
`pipx install voxtype`.

Note: voxtype only works on macOS (it types via macOS accessibility
APIs). Tell the user this up front if they are on Linux/Windows.

## Configure

Run the interactive wizard — it asks a handful of explained questions
(engine, activation mode, hotkey, ...) and writes
`~/.config/voxtype/config.toml`:

```bash
voxtype setup
```

This must run in the user's own terminal (it is interactive); do not
run it through an agent shell. Alternative for hand-editors:
`voxtype init` writes a fully commented sample config instead.

### Editing the config directly

All settings live in `~/.config/voxtype/config.toml` — every key is
optional and the file `voxtype init` writes is fully commented. You (the
agent) can open and edit this file to help the user tune specific
options; common ones:

- `engine` — `parakeet-mlx` (Apple GPU), `parakeet` (CPU), or `moonshine`.
- `mode` — `toggle` (hotkey), `wake` (wake word), or `vad` (always on).
- `segmentation` — `hold` (whole take on toggle-off) or `vad` (per pause).
- `hotkey` — toggle chord, e.g. `"<ctrl>+;"` (run `voxtype hotkey` to
  record one).
- `sounds` — set `false` to silence the start/stop chimes.
- `wake_word` / `wake_word_aliases`, `submit_phrases`, `overlay`,
  `paste_hotkey`, `cancel_hotkey`, `copy_to_clipboard`, and more.

For the full list — every key, its default, and the engine comparison —
read the **Configuration & CLI** reference:
<https://pchalasani.github.io/claude-code-tools/tools/voxtype/configuration/>

## Launch

```bash
voxtype
```

Then: press `Ctrl+;` (default hotkey), speak, press `Ctrl+;` again —
the take is transcribed and typed at the cursor. `Esc` cancels a
recording. Saying "go", "over", or "submit" alone presses Enter.

First-run notes for the user:

- macOS will prompt for **Microphone**, **Accessibility**, and
  **Input Monitoring** permissions for the terminal app — all three
  must be granted, then voxtype restarted.
- The first launch downloads the speech model (up to ~2 GB for the
  default Apple-GPU engine, ~490 MB for the CPU engine); later launches
  skip the download and just load it onto the GPU (a few seconds, shown
  by a spinner).

## Reference

- Config file: `~/.config/voxtype/config.toml` (fully commented; edit to
  tune any setting)
- `voxtype --help` — all flags and subcommands
- Config reference (every key + defaults):
  https://pchalasani.github.io/claude-code-tools/tools/voxtype/configuration/
- Overview & full docs:
  https://pchalasani.github.io/claude-code-tools/tools/voxtype/