Settings

Wisspa's settings live in eleven tabs. Open them from the pill (click it → Settings) or with the app focused. Everything saves immediately; nothing needs a restart.

General

Day-to-day behaviour: launch on login, show/hide the pill, recording mode (push-and-hold vs toggle), sounds and volume, the ready chime and dictation-complete sound, mic sensitivity, input device, max recording length (10–120 s), Re-calibrate (re-runs the two-sample mic calibration from onboarding), quiet notifications, verbose logging, and automatic update checks.

API Keys

Your Groq (speech-to-text) and Anthropic (cleanup + prompts) keys. Stored in the macOS Keychain. Each has a Save & test button that verifies the key against the provider's API so typos surface immediately.

Hotkeys

Reassign all five actions — see the Hotkeys doc.

Actions

The voice-action registry: all loaded actions with their triggers, types, and enabled state, plus a shortcut to open the actions folder in Finder. Editing happens in the YAML files — see the YAML reference.

Prompt Mode

The rewriter's knobs: include selected text as context, preview before insert with a countdown slider, review before insert (approve every prompt in a window first), manual app override for when app detection misfires, a free-text user profile injected as standing preferences ("I'm a contract lawyer, plain English please"), and adaptive refine — the two-pass critique-and-revise for complex asks, on by default.

Corrections

Word corrections Wisspa has learned from your edits — fix "furseller" to "Vercel" a couple of times and it starts auto-applying. Set the learning threshold, toggle auto-apply, or edit/remove individual entries.

Profiles

Per-app tone + vocabulary for dictation. Match an app by name (e.g. "Slack"), give it a tone note ("casual, no greetings") and extra vocabulary. Profile vocab wins over global entries, and profiles feed the transcriber's hint first.

Vocabulary

Global custom words — names, jargon, product spellings — sent to the transcriber as hints. Add one by one or bulk-import a CSV (spoken,replacement); the importer previews additions and skips duplicates and bad rows.

History

Every dictation, action, and prompt: the raw transcript, the output, the app it went to, latency, token usage, and status. Searchable, filterable by mode, re-injectable with one click, exportable to CSV, clearable, and auto-pruned at 1000 rows. All local — history.db in Application Support.

About

Version, the live permissions diagnostic (all four grants with deep-link buttons into System Settings), Export Diagnostics (a redacted .zip for bug reports), and update status.