7.2 KiB
Settings by tab
Open the microphone icon in the Desktop tray and choose Configure… to open Frame Voice Configuration.
You can make changes across all three tabs, then click Save settings once in the shared footer. Save is disabled until something changes. Saving applies the settings and restarts dictation, so finish any recording first.
General
| Setting | What it does |
|---|---|
| Text delivery | Typing sends keystrokes; Paste uses the clipboard in compatible standalone apps. Desktop uses typing automatically. |
| Start and stop tones | Enables or disables start, stop, and cancel feedback sounds. |
| Separator after dictation | Space keeps consecutive dictations from running together. None adds no separator. |
| Maximum recording seconds | Limits each recording; the default is 30 seconds. |
The tray shows a white microphone while dictation runs and a white crossed-out microphone when paused. Use the tray's start/resume or pause options to control it. In the headset, the small microphone waits for audio startup, then grows toward the start tone. Wait for the tone before speaking. Release the joystick cap to finish; click it during recording to cancel.
Updates
The Updates group shows your installed version. Development builds also show a build revision so you can distinguish versions tested between releases.
Click Check for updates to compare it with the latest stable release. Internet access is required. Checks run only when you request them.
If an update is available, save any pending settings, then choose Update… to open the installer. Updates retain saved settings, API keys, and downloaded models. Open releases lets you read the release notes. Experimental releases are not selected by this check, and newer installed versions are not downgraded.
Uninstall
Uninstall… is at the bottom of General. It opens a confirmation dialog with both retention options checked by default:
| Confirmation option | Checked | Unchecked |
|---|---|---|
| Keep current configuration | Keeps saved settings, API keys, and your managed custom prompt. | Removes them. |
| Keep downloaded models | Keeps Frame Voice's model downloads for later reuse. | Deletes them to recover storage. |
Uncheck both to remove Frame Voice's settings, saved keys, downloaded models, and app data. Installation backups are removed during uninstall. Separately installed voice tools and their models are left alone.
Transcription
Transcription path
| Model | Runs on | Speed | Accuracy | Internet needed | Requirements |
|---|---|---|---|---|---|
| Whisper Base | Local | Medium | Iffy | Initial download only | ~142MiB |
| Whisper Small | Local | Slow | Good | Initial download only | ~466MiB |
| Whisper Large v3 Turbo | Groq | Fast | Best | Every dictation | API key |
Whisper models were created by OpenAI. Local transcription sends no audio to OpenAI or Groq; the models run on your Frame using whisper.cpp.
Select a path, then Save settings. Local transcription works offline after its initial download. Switching to local retains your saved Groq API key so you can switch back later. Existing verified model downloads are reused.
If downloading or preparing a new model fails, the current transcription path is retained.
Language
Choose a language from the searchable list, or Automatic detection when you use more than one language. English local transcription uses an English model; other languages and automatic detection use a multilingual model. Changing this selection may require another download.
For accented characters and non-Latin scripts, use a clipboard-compatible standalone app such as Firefox. Typing inside Desktop currently supports ASCII. Controller buttons work independently of the transcription language; spoken submit/clear command aliases remain English.
Groq options
These controls appear when Groq · Cloud is selected:
| Setting | What you need to know |
|---|---|
| Groq API key | Paste your key here. Get a Groq API key opens Groq's key page. The key stays saved when switching to local transcription. |
| Clean up transcript | Removes fillers and stutters, resolves self-corrections, and adds punctuation using Groq. |
| In-headset notifications | Controls Groq usage and limit notifications. |
| Convert spoken slash to / | Optionally turns a spoken “slash” into / during cleanup. It is off by default. |
Edit cleanup prompt… opens the built-in editor. Customize the instructions, click Apply, then Save settings. Apply updates the draft; Save writes it. Reset to default restores the shipped prompt in the editor. Apply and Save that draft to keep the reset. Cancel discards edits made inside the editor.
The default prompt asks for neutral cleanup while preserving meaning, language, names, and technical text. Your custom prompt stays on your Frame and is preserved by updates.
Quiet recordings and uncertain Groq results may cancel instead of inserting text. This reduces invented transcripts from background sounds.
Downloaded models and storage
The Downloaded models section shows the stored models and their sizes. English and multilingual copies appear separately.
To free space without uninstalling, click Delete… beside an unused model and confirm. The active model is protected: select and save another transcription path or model first. Choosing a deleted model later downloads it again.
Controllers
Enable dictation for each hand and choose the submit and clear buttons. Button names are shown as A, B, X, Y, or D-pad Up/Down/Left/Right. Choose None to disable a particular action. Save to apply new bindings.
| Default control | Action |
|---|---|
| Either joystick cap | Touch briefly, release, then touch and hold to record. |
| Joystick click during recording | Cancel the recording. |
| Right A / left D-pad Right | Submit after text appears. |
| Right X / left D-pad Left | Clear after text appears, when the button is released. |
Submit and clear are available for ten seconds after successful dictation and stop when focus changes. A plain joystick hold does not start recording.
B's Desktop space
The default Clear button is X, which does not type an extra character. If you choose B instead, Steam's default Desktop layout also types a space. Leave B also types a space in Desktop enabled to include that extra character when clearing. Disable it if your custom Steam B mapping does not type a space. This adjustment applies to right-hand B in Desktop; spoken Clear and other bindings keep their own behavior.
Desktop Clear tracks the text inserted by the current Frame Voice process, including its separator. Restarting dictation resets that history. Test Clear with a fresh dictation after an update or settings save. Moving the cursor or editing the text yourself can also invalidate the tracked count.
More detail
See troubleshooting if dictation is not working, or uninstalling Frame Voice for removal and retention choices. Advanced configuration belongs in the developer guide.