Download Dictate

Sign in, hit record, and your speech appears as text in real time. Audio is relayed frame-by-frame and never written to disk.

macOS

Apple silicon & Intel

Universal build (DMG). Requires macOS 11 or later.

Download for macOS (.dmg)

• Apple silicon (M1/M2/M3/M4)

• Intel (x64)

• Also available as a portable .zip

Windows

64-bit

NSIS installer. Requires Windows 10 or later.

Download for Windows (.exe)

• Windows 10 / 11 (64-bit)

• Portable .zip also available

• Choose the install location during setup

Build from source

The client is open source. You need Node.js 18+ and a Dictate account. 

cd client
npm install
npm start          # run in development
npm run build:mac  # package for macOS
npm run build:win  # package for Windows
npm test           # run the client test suite

How it works

1

Sign in

Authenticate against your Dictate account.

2

Get a relay token

A 5-minute token scoped to your account and quota.

3

Record

16 kHz mono PCM in ~200 ms frames, relayed untouched.

4

See the text

Recognition results stream back and appear on screen.

Need help? support@dictate.top