Documentation

Everything you need to use Dictify.

Getting started

Installation

Download Dictify from your dashboard. Open the DMG, drag Dictify to Applications, and launch it. On first run, Dictify installs its dependencies and downloads the Whisper model automatically. This takes about a minute.

Permissions

Dictify needs two macOS permissions. You'll be prompted on first launch:

Accessibility — to detect your hotkey and insert text at the cursor.
Microphone — to record your speech.

Grant both in System Settings > Privacy & Security.

How to dictate

Double-tap Ctrl (or your configured hotkey).
Keep holding after the second tap and speak.
Release to stop recording. Your text appears at the cursor.

Features

Push-to-talk

Hold your hotkey to record, release to transcribe. Dictify uses a double-tap to start — tap the trigger key twice quickly, hold on the second tap, speak, release.

Punctuation commands

Say "period", "comma", "question mark", "new line", or "exclamation mark" and they'll be converted to symbols. Sentences are auto-capitalized after periods. Programming punctuation is also supported: say "open brace", "arrow right", "double equals", and more. German commands like "Punkt", "Komma", and "Fragezeichen" work when your language is set to German.

Voice Activity Detection

Dictify automatically detects silence and skips transcription when you don't speak. Short recordings under 0.5 seconds are discarded.

Recording indicator

A small animated pill appears at the bottom of your screen while recording. You'll also hear a sound when recording starts and stops.

28 languages

Set your language from the dashboard. Dictify supports English, German, French, Spanish, Italian, Portuguese, Dutch, Polish, Russian, Chinese, Japanese, Korean, and 16 more. Changes take effect after restarting the app.

Configurable hotkey

Choose your hotkey from the dashboard: double-tap Ctrl, Option, or Cmd, or use a key combo like Option+S or Ctrl+Shift+Space. Changes take effect after restarting.

AI features

Beyond transcription, Dictify can translate your speech, summarize long dictations, fix grammar, or rewrite text in different tones. Access these from the menu bar under the AI menu.

Security

Two-factor authentication

Protect your account with TOTP-based 2FA. Go to Settingsand click "Set up 2FA". Scan the QR code with any authenticator app (Google Authenticator, Authy, 1Password). You'll need a code every time you sign in.

Session management

View all devices signed in to your account in Settings. Revoke individual sessions or sign out everywhere with one click. Each session shows the browser, IP address, and last active time.

Privacy

Dictify processes all audio locally on your machine by default. No audio data is ever sent to our servers. Your transcriptions are yours. If you use cloud-whisper, audio is sent to your self-hosted server only.

Configuration

Dashboard settings

Your dashboard lets you configure the hotkey and language. The app syncs these settings from the server on each launch.

Whisper models

Dictify uses OpenAI's Whisper for transcription. Available models:

Model	Size	Speed	Accuracy
tiny	75 MB	Fastest	Low
base	139 MB	Fast	Good
small	461 MB	Medium	Very good
turbo	1.5 GB	Medium	Excellent
large-v3	2.9 GB	Slow	Best

The default is small — a good balance of speed and accuracy. Change it in ~/.config/glace/config.json.

Local config file

All settings are stored at ~/.config/glace/config.json. Dashboard changes (hotkey, language) are synced on app launch. You can also edit the file directly to change the model or other advanced settings.

Transcription backends

openai-whisper — Default. Works on all platforms. Supports all models.
faster-whisper — 2-4x faster on CPU using CTranslate2. Best for Intel Macs and Linux.
mlx-whisper — Optimized for Apple Silicon. Fastest on M1/M2/M3/M4.
cloud-whisper — Offload to your own server. See our self-hosting guide.

API reference

License validation

The desktop app validates its license key on startup:

POST /api/license/validate
Content-Type: application/json

{ "license_key": "XXXXX-XXXXX-XXXXX-XXXXX" }

Returns the user's plan, quota, and validation status.

Usage tracking

POST /api/usage
Content-Type: application/json

{ "minutes": 1.5 }

Reports transcription usage. Free plan users are limited to 10 minutes per day.

Full API spec

The complete OpenAPI 3.1 specification is available at /openapi.json.

Troubleshooting

App doesn't start

Dictify requires Python 3.10+ installed via Homebrew. If you don't have it, run brew install python in Terminal.

Hotkey doesn't work

Make sure Accessibility permission is granted for Python in System Settings > Privacy & Security > Accessibility. You may need to remove and re-add it.

Transcription is wrong language

Set your language in the dashboard and restart Dictify. The default is English. Auto-detect works poorly on short recordings.

Slow transcription

Try a smaller model (tiny or base) for faster results. On Apple Silicon, switch to the mlx-whisper backend for significantly faster performance. Or set up a cloud Whisper server.

Where are the logs?

Logs are at ~/.local/share/glace/launcher.log. Run tail -f ~/.local/share/glace/launcher.log to watch in real time.