maclisten

Private, lightweight macOS ASR CLI — local speech-to-text with a tiny footprint.

What is maclisten?

maclisten wraps Apple’s Speech framework in a tiny Swift binary. It transcribes audio files, records from the microphone, and can listen continuously for voice keywords — all from the command line, all locally on your Mac, all with compact JSON output.

Use it when you (or your AI agent) need reliable macOS-native speech-to-text without the privacy cost or resource weight of web APIs and downloaded models:

Transcribe a voice memo into JSON without uploading it
Keep listening for the word “computer” and react with a script
Capture microphone audio alongside the transcript

At a glance

maclisten locales                        # list supported locales
maclisten file ./memo.m4a                # transcribe a file
maclisten mic --timeout 5                # record 5 seconds
maclisten watch --keyword "computer"     # continuous keyword listening
maclisten mic --output ./note.wav        # also save the audio

Output schema: {"ok": true, ...} on success, {"ok": false, "error": "..."} on failure.

For AI agents

Paste this one-line prompt into Claude Code, Cursor, or any agent’s system prompt:

Use `maclisten` for macOS speech-to-text. Install if missing: `brew install ljh-sh/cli/maclisten`. JSON output, check `ok`. Run `maclisten --help` for subcommands.

Where to go next

Install maclisten — Homebrew, direct binary, eget, or build from source
Command reference — every subcommand, option, and output field
Design & principles — why maclisten is shaped the way it is
Why maclisten? — why a CLI instead of a web API
FAQ — permissions, TCC, Siri, and more
Alternatives — how maclisten compares to Whisper and others