S

Speech-to-Markdown

A Mac menu bar voice app that transcribes locally and turns speech into clean Markdown documents in real time.

Last Updated 9/11/2026

About Speech-to-Markdown

Speech-to-Markdown is a Mac menu bar app that handles transcription and organization on-device. In Global Dictation mode, press the shortcut in any app and speak; the text is entered directly at the cursor. Agent Mode uses a local model you run to organize spoken content into structured documents in real time, supporting Markdown, plain text, and HTML, with the raw transcript available in one click. Audio is transcribed on-device by whisper.cpp, and the local LLM server must support the OpenAI request format. An iOS version also offers the same Agent Mode 100% offline on iOS 26+ and iPhone 15 Pro or newer.

Best for

Writers and note-takers who need voice capture
Developers who often enter text by voice in terminals, browsers, and similar contexts
Mac users who value privacy and want data to stay on-device
Mobile users who record offline on iPhone or iPad

Best use cases

Dictate into a terminal, browser, or chat app, with text entered directly at the cursor

Turn spoken ideas into structured Markdown documents in real time

Keep the raw transcript after dictation and view it in one click when needed

Record voice completely offline on iPhone or iPad

Features

Global Dictation

Press ⌘⌥] in any app and speak. Transcribed text is entered directly at the cursor, with a Spotlight-like status bar showing the current state.

Agent Mode Real-Time Organization

Speak freely while a local model organizes your words into structured documents in real time. Supports Markdown, plain text, and HTML, with the raw transcript available in one click.

On-Device Transcription

Audio is transcribed on-device by whisper.cpp in segments of about 4 seconds. Agent Mode connects to your local LLM server, with the default http://127.0.0.1:8000/v1.

Offline iOS / iPadOS Version

Use the same Agent Mode 100% offline on iOS 26+ and iPhone 15 Pro or newer. SpeechAnalyzer transcribes the audio and Apple Intelligence organizes the document.

Pros

  • Transcription and document organization happen on-device; the iOS version is also 100% offline
  • Global dictation enters text directly at the cursor in any app
  • Supports Markdown, plain text, and HTML output
  • Install with a one-line script; whisper.cpp is installed automatically

Cons

  • Agent Mode requires you to run a local LLM server on your Mac
  • The iOS version requires iOS 26 or later, an iPhone 15 Pro or newer, and a device that supports Apple Intelligence

FAQ

Everything runs on-device: audio is transcribed by whisper.cpp on the device, and documents are organized by the local model you run on your Mac; the iOS version is also 100% offline.
It connects to the local LLM server running on your Mac. The default address is http://127.0.0.1:8000/v1, and the server must support the OpenAI request format.

Tool Information

DeveloperSpeech-to-Markdown
Release Date9/11/2026
PricingFree