Skip to main content

​
Works in any app

Most dictation tools only work in their own window. You dictate, then copy-paste. Mumbli skips that entirely — it types directly where your cursor is.

​
No plugins. No integrations.

Mumbli uses the macOS Accessibility API to inject text at your cursor position. This means it works in every app that has a text field, without any app-specific setup:

Browsers

Chrome, Safari, Firefox, Arc — forms, Gmail, Google Docs, any web app

Email

Mail, Outlook, Spark — compose emails by speaking

Messaging

Slack, Discord, Messages, Telegram, WhatsApp — reply without typing

IDEs

VS Code, Xcode, IntelliJ, Cursor — dictate comments, docs, or code

Notes

Notes, Obsidian, Notion, Bear — brain dump by voice

Terminal

Terminal, iTerm, Warp — even command lines work

​
How it works under the hood

  1. You activate Mumbli (hold Fn or double-tap)
  2. You speak — audio is transcribed and polished
  3. Mumbli finds the focused text field using AXUIElement (Accessibility API)
  4. Text is injected at the cursor position
  5. If the Accessibility API can’t reach the text field, Mumbli falls back to clipboard paste (Cmd+V)
The fallback ensures dictation works even in apps with non-standard text fields.

​
The cursor stays where you put it

Mumbli doesn’t open a new window, steal focus, or move your cursor. You click where you want text, hold Fn, speak, and the text appears right there. Your workflow doesn’t change — you just stop typing and start talking.

​
Permissions

For this to work, Mumbli needs Accessibility permission in macOS: System Settings > Privacy & Security > Accessibility > Mumbli This gives Mumbli write-only access to text fields. It doesn’t read your screen or monitor what you type — it only injects the text you dictated.

Download Mumbli

Free and open source. Works in every app on macOS 13.0+.