Dictation modes
The Dictation page is where you create and tune profiles (modes) that control how Kalam hears you and how it writes. Star a mode as your default, open a card to edit its pipeline and auto-switch rules, and use While dictating for settings that apply to every mode. Concepts behind modes: Modes.
What you see
Header
| Element | Purpose |
|---|---|
| Dictation title | Main modes screen |
| Hold / cycle hotkey hints | Reminds you which keys start dictation and cycle modes |
| New mode | Creates a custom profile and opens its editor |
Your modes
Each card is one profile:
| Element | What it means |
|---|---|
| Icon / logo | How the mode appears in the list, status bar, and pill |
| Name | Profile label (built-in Base name is locked) |
| Default badge | Starred mode used when you dictate unless another rule or switch applies |
| Star button | Sets this mode as default (disabled on the current default) |
| Feature chips | AI polish, Context, Auto-switch (or Auto-switch ×N), plus Built-in on Base |
| Plain transcription | Shown when no polish, context, or auto-switch is on |
| Health alerts | Red messages when voice or AI setup is incomplete (missing key, engine, or plan) |
Star sets the durable home mode. Click the card body (not the star) to open the editor.
Mode editor
Opened from a card. Sections:
- Identity — name, icon or brand logo, accent color (how the profile looks in the list, status bar, and pill).
- What happens when you dictate — the pipeline rail (see below).
- Auto-activate — rules that switch to this profile when a matching app or site is focused.
In the detail view you also get Set as default (or a Default mode pill) and, for deletable modes, Delete mode.
Pipeline rail
Heading: What happens when you dictate — “Follow your words from microphone to typed text.”
Flow: You speak → three steps → Typed into your app.
| Step | Title | What you control |
|---|---|---|
| 1 | Hear your voice | Voice recognition provider/model (or local engine) and recognition languages |
| 2 | Read your surroundings | Optional context from your computer |
| 3 | Improve the text | Improve with AI, AI model, and How you want it written |
Toggle Show details / Simple view for expert controls (providers, models, languages). Non-Base modes can use Same as Base for voice, AI, and languages.
Read your surroundings (master: Read surroundings):
| Context source | Options |
|---|---|
| Active app content | Off, Window title, Focused field, Extended |
| Clipboard | On/off — last thing you copied |
| Selected text | On/off — highlighted text when you start dictating |
| Date and time | On/off |
Improve the text — master toggle Improve with AI. When on: choose provider/model and optional writing instructions. Built-in modes can Reset to default if you changed the prompt.
If voice stays on-device but improvement uses a cloud AI, the rail shows a privacy note that text improvement still sends data to that provider.
Pipeline deep dives: Pipeline overview, STT modes, Context awareness, AI polish.
While dictating
Global section under the mode grid — “Applies to every dictation mode and profile.”
| Control | What it does |
|---|---|
| Pause music and video | Pauses Spotify, Apple Music, and browser video when possible |
| Voice commands | Spoken commands like “scratch that” / “delete that” when Improve is off. Not the same as voice actions (new note, press enter). |
| Quiet other apps | Lowers volume of other apps (not on macOS in this section; on Windows it may show as coming soon / disabled) |
Restore banner
If you removed shipped profiles, a banner may say N built-in modes not in your list with Restore all — restores profiles like Email, Coding, or Notes.
Common tasks
Create a mode
- Open Dictation in the sidebar.
- Click New mode.
- Kalam creates a profile named New mode (Improve on by default; surroundings off; inherits Base voice/AI) and opens the editor.
- Set Identity, tune the pipeline, then add Auto-activate rules if you want.
Set the default mode
- On Your modes, click the star on a card that is not already default, or
- Open the mode and click Set as default.
The default shows a Default badge and a filled star.
Edit how a mode hears and writes
- Click a mode card.
- In What happens when you dictate, adjust Hear your voice, Read your surroundings, and Improve the text.
- Use Show details when you need provider, model, or language lists.
- Changes save as you edit (leave fields / toggles as usual).
Add an auto-switch rule
- Open a mode → expand Auto-activate.
- Click Add Rule.
- In Add Auto-activation Rule, pick an open window or browse apps. For browsers you can match websites (domains).
- Confirm — the rule applies to all windows of that app (not only the title you saw), unless you chose a website rule.
Conflicts: if two modes claim the same app, the editor warns which one wins. Full rule types: Modes.
Restore built-in modes
- If the restore banner appears, click Restore all.
- Wait until it finishes (Restoring…), then confirm the modes are back in Your modes.
Delete a mode
- Open a deletable mode (Base cannot be deleted; other built-ins may be removable).
- Click Delete mode (or Delete in the panel footer).
- Confirm when asked.
Deleted built-ins can be brought back with Restore all.
Plan gates
| Feature | Free / Expired | Trial / Pro / Max |
|---|---|---|
| Create & edit modes, auto-switch rules | Yes | Yes |
| Improve with AI & Read surroundings | Locked (PRO) | Editable |
| Pause music and video | Locked (PRO) | Available |
| Voice commands | Available | Available |
| Quiet other apps | Locked when shown | May still be disabled if not shipping yet |
Upgrade or sign in from Account. Max also unlocks Kalam-hosted cloud AI where offered — see AI and models and Choosing an STT mode.