Use Freehand

Choose a task from the activity rail on the left. Each task has its own connection and model settings; only Voice needs a microphone. If you have not configured a service yet, start with Get started.

I want to…	Open	Result
Dictate into another application	Voice transcription	Text inserted while the captured destination is still safe, or available to copy
Transcribe an existing recording	Audio file	A transcript you copy explicitly
Hear text read aloud	Text to speech	Audio you can play and explicitly save

Press Ctrl + K on Windows or ⌘ + K on macOS to find a task or settings page. The rail also opens Connections, Local runtime, History, and Settings.

Voice dictation

Complete the Voice readiness screen with a usable microphone and transcription service. A recording shortcut is optional, but lets you begin with another app’s text field focused. See first dictation setup.

1. Focus the destination text field. Keep this application and window open.
2. Start recording with your toggle or hold-to-talk shortcut. Wait for Freehand to show that recording has started, then speak.
3. Stop recording with the toggle shortcut, or release the held chord.
4. Keep the destination focused while transcription and any enabled cleanup finish. The completed text appears there when delivery is safe.

Check before pasting a recovery result

If delivery stops, use Copy on the current result. Some text may already have been inserted: check the destination before pasting to avoid duplicates. Windows requires the original focused control; macOS allows another field in the same app and window. See safe insertion.

To change inputs, choose Microphone at the top of the Voice sidebar. The selection saves immediately for the next recording. System default microphone follows the system setting; Refresh microphones finds a newly connected device. A disconnected saved device stays selected until you reconnect or replace it.

Toggle recording

Press Ctrl + Shift + Space to start and press it again to stop, unless you have changed the shortcut. With voice detection and automatic stop enabled, a pause after speech can also finish the recording.

You can Clear the toggle in Settings → Shortcuts and use on-screen controls instead. Starting in Freehand may require copying the result because the intended destination was not focused at recording start.

Hold to talk

Assign a chord in Settings → Shortcuts, hold it while speaking, then release it to finish. Automatic silence ending does not stop a held recording. Supported modifier-only combinations and conflict recovery are covered in Keyboard shortcuts.

Long dictation and VAD

Voice activity detection (VAD) identifies speech and silence on your computer and is on by default. These optional controls in Voice transcription → Audio start off:

Control	What changes
Trim silence	Removes silence from the beginning and end
Stop after I finish speaking	Ends a toggle recording after the configured pause
Split long dictation on silence	Transcribes checkpoints while you continue; prefers pauses and can split at the duration limit

Checkpoint transcripts are assembled and delivered together when you finish. For a preview that changes while you speak, use Live transcription.

Audio files

Open Audio file, choose a recording, then choose Transcribe. File selection does not upload anything until you start. Results always need explicit Copy; you do not need a microphone or completed Voice setup.

Transcribe an audio fileChoose a file, use streamed results, and recover from an incomplete request.

Text to speech

Open Text to speech to generate audio from your own text. After enabling a speech connection, model, and voice, choose Speak. You can also Listen to a completed transcript, with history on or off.

Generate and play speechSet up a voice, preview it, control playback, and save generated audio.

Transcript cleanup

Optional Cleanup, shared by Voice and audio files, sends the raw transcript to a separate chat service. Keep it off for unchanged text, use custom instructions, or choose S1-mini’s English cleanup controls. A failure, timeout, or empty response falls back to raw text.

Set up transcript cleanupChoose a processor, understand language restrictions, and compare results.

History

History is off by default. Enable it before a run to keep both raw and cleaned text for comparison. It holds at most 20 entries and 2 MiB in memory, and clears when Freehand exits. Copy text you want to keep before quitting.

Browse session historyFind, compare, copy, or remove retained transcripts.

Current result and task settings

Current result holds the latest transcript even with history off. Copy recovers a result after failed insertion; Clear removes it; Listen plays it when speech is enabled. Starting another recording replaces the Voice result. Clear or replace a selected file to remove its file result.

Select a row in the task sidebar, or its Settings cog, to open detailed options beside the result. Quick controls save immediately; detailed edits use Save. Changing settings does not alter work already running.

Arrange your workspaceSidebars, panels, window controls, connection checks, and CPU / RAM / GPU readings.

Manage saved connectionsEdit server addresses and credentials, then select them for each task.

Finding and adjusting settings

Use Settings for General, Shortcuts, Overlay, and Vocabulary. Use each task’s Settings cog for transcription, audio, cleanup, speech, or delivery options. Find settings searches by page name or terms such as “glow” and “hotkey”; Enter opens the first match and Escape clears the filter.

Leaving unsaved options offers Save, Discard, or Keep editing. A failed save keeps the draft and previous applied settings. The workspace guide explains quick controls, detailed options, and model pickers.

Action feedback

Open Details beside an error to see the explanation and relevant settings. After correcting the problem:

Failed action	Recovery
Voice recording	Record again; discarded microphone audio cannot be retried
Audio file	Retry the selected file
Speech composer	Try again with the text currently in the editor; the draft is preserved

See Troubleshooting for the visible failure category.

Status overlay and tray

The passive overlay shows recording, processing, and copy-required state without taking focus. Error and needs copy notices disappear after five seconds; the result or error remains available in Freehand. Configure it under Settings → Overlay, or add live captions.

Click the Windows tray or macOS menu-bar icon for recording controls, the Voice connection and model, and the latest result with Copy. Choose Open Freehand for the full workspace. Escape or clicking outside dismisses the panel without cancelling work. On Windows, focus the destination and use your shortcut when you want to choose exactly where recording will deliver.

Closing the window keeps Freehand running

Resolve any unsaved settings, then close to hide the workspace. Right-click the tray or menu-bar icon and choose Quit to exit and cancel active work. Let an audio export finish first: interrupting it can leave a partial WAV file.

Appearance

Choose System, Light, or Dark under Settings → General. Windows also offers Mica. Native appearance changes take effect after quitting from the tray and reopening Freehand.