---
title: 'Voice Dictation for AI Prompts on macOS'
excerpt: 'Speak your AI prompts instead of typing them: hold a hotkey, say the whole request, and Keebye types it into any tool — on-device and private.'
date: '2026-07-15'
author: 'Teodor Deleanu'
tags: ['ai-prompts', 'dictation', 'voice-to-text']
faq:
  - question: 'Which AI tools does this work with?'
    answer: 'All of them. Keebye types into whatever field has focus, so ChatGPT in the browser, Claude, a coding assistant, or any chat box gets the same treatment — there is no per-tool plugin to install.'
  - question: 'Does the AI provider get my audio?'
    answer: 'No. The speech-to-text step happens on your Mac before a single character reaches the tool. The model on the other end receives plain text and cannot tell it came from your voice.'
  - question: 'Is spoken text good enough to send as a prompt?'
    answer: 'Usually, yes. Rule-based cleanup fixes the obvious spoken-language artifacts, and an optional on-device polish pass tightens it further while guarding against changing your meaning. You still see it before you hit send.'
  - question: 'How long a prompt can I dictate at once?'
    answer: 'As long as you like. Tap the hotkey once to toggle dictation on for a multi-paragraph brief, speak it out, then tap again to finish — you are not limited to a quick hold.'
  - question: 'What about code, file names, or library names in my prompt?'
    answer: 'Teach them to the custom dictionary once. Repo names, package names, and API terms then transcribe correctly, so the prompt reads the way you meant it without a cleanup pass.'
keywords: ['voice to text for ai', 'dictation for ai prompts', 'speech to text for ai chat']
---

## Why speak your prompts

A good prompt is a paragraph, not a search query. You are describing context, constraints, the shape of the output you want, the edge cases to avoid — and typing all of that is the slow part of working with AI. The thinking is fast; the transcribing of the thought is what drags.

Say it instead. Speaking a multi-sentence request is closer to how the idea actually arrives in your head, and it comes out in one breath rather than one keystroke at a time. This is the heart of what Keebye is for: your voice becomes the input to every AI tool you touch, whether that is [Claude Code in the terminal](/dictation-for/claude-code) or a chat window in the browser.

## How it works

Click into the prompt box, hold Right ⌘, and describe what you want the model to do. Let go and the full request appears at the cursor, ready to send or refine. For a longer brief, tap the key once to keep dictation open, talk through the whole thing, and tap again to stop.

The transcription is on-device — Parakeet, the English default, downloads once and then runs with no connection at all. Nothing you dictate is logged or uploaded, so even a prompt full of proprietary context never leaves your Mac before the tool sees it as ordinary text.

The identical hotkey works in [ChatGPT in your browser](/dictation-for/chatgpt) and in a native app the same way, so switching tools costs you no new habit.

## Setup in two minutes

Allow Accessibility and microphone access on first launch, then pick your trigger key — Right ⌘, Fn, or Right ⌥. Enable launch-at-login and Keebye is waiting the next time a prompt box is in front of you.

Spend the last minute on the custom dictionary. Prompts are dense with proper nouns — model names, product names, the libraries you reference constantly — so teaching those once means your requests come out clean instead of needing edits before you send.

## Limits, honestly

Keebye writes the prompt; it does not run it. There is no integration that reads a tool's state or clicks send for you — it drops text where your cursor sits and you stay in control of the conversation, the same as you would dictating into [Cursor's composer](/dictation-for/cursor).

The text arrives when you release or toggle off the hotkey, not word by word as you talk — transcription is batch, not streaming. And Keebye is macOS only.