Configuration
Every global HoldSpeak setting. Changes apply to your local runtime.
Fix these before saving
Pins transcription for dictation, meetings, and import. Auto-detect works well for longer speech; pin your language if short utterances get misdetected.
Your own spoken words for symbols, merged over the built-in punctuation commands (yours win). Say the phrase while dictating and the symbol is typed. Attach controls spacing: none keeps it, left/right/both glue the symbol to its neighbors.
No custom symbols yet. The built-in commands (period, comma, dash, new line) always work.
The openWakeWord model name. "hey_jarvis" ships pretrained.
Preview shows the result with a Type it button; nothing is typed until you confirm. Type immediately is your call to make: a false detection would type into whatever app is focused.
Turn on Desktop presence too: it is the always-visible armed indicator while you work in other apps.
Adds a Send to Slack button to each meeting's aftercare. Sends exactly what you preview, after you approve each send, only to this URL's host. Leave empty to turn the feature off.
These apply only when intel provider is cloud or auto — the only settings that can send a transcript off your machine.
cloud
auto