Voice Commands

Voice Commands let you navigate Predite hands-free. Tap the microphone button, say where you want to go — "open scanner", "show whales", "go to portfolio" — and Predite jumps you there instantly. It is a navigation accelerator built for traders who keep one hand on a chart and the other on a coffee, and who do not want to break focus hunting through a sidebar when a market is moving.

This is not a chatbot and it does not place trades. Voice Commands map spoken phrases to dashboard destinations. Think of it as a spoken command palette: faster than clicking, available in four languages, and processed entirely inside your browser.

## What It Does

Voice Commands recognize short navigation phrases and route you to the matching page in one step. Instead of moving the cursor to the sidebar, scanning the labels, and clicking, you say two or three words and you are there.

The feature covers the destinations you reach most often during a session:

Matching is **partial and case-insensitive**, so you do not need to memorize exact phrasing. Any utterance that *contains* a known keyword wins. "Take me to the scanner please" matches `scanner` and routes you correctly. "Let's check the whales" matches `whales`. You can speak naturally; Predite picks out the keyword.

When a command matches, you get a confirmation toast showing the transcript and the destination, then the page changes. When nothing matches, you get a toast telling you the command was not recognized, and you stay where you are — no surprise navigation.

## How to Use It

1. **Find the mic button.** On desktop, look for the round microphone button in the bottom-right corner of the screen, stacked just above the floating AI Chat widget. It sits on top of every dashboard page. 2. **Tap it once.** The button turns red and pulses to show it is listening. Your browser may ask for microphone permission the first time — click **Allow**. 3. **Speak a command.** Say a short phrase like "open scanner" or "show whales". One phrase per tap. You do not need to hold the button down. 4. **Watch the confirmation.** A toast appears with what Predite heard and where it is sending you, then the page loads. If the command was not recognized, the toast says so and nothing changes. 5. **Repeat as needed.** Tap again for the next command. The mic stops automatically after each utterance, so you are never left with a hot microphone.

To cancel mid-listen, tap the button again (now showing the red recording state). That aborts the current capture silently, with no error and no navigation.

### Tips for Reliable Recognition

## Supported Languages

Voice Commands recognize speech in **four languages**, and the recognizer automatically follows your Predite interface language:

You do not pick a voice language separately. Whatever language you have set for the Predite UI is the language the microphone listens in. If your interface is in Portuguese, the recognizer expects Portuguese and the Portuguese keywords ("baleias", "carteira", "escanear") are active. Switch the UI to Spanish and it switches with you on your next command.

Because the keyword map includes both English and Portuguese aliases for most destinations, English and Portuguese speakers have the richest vocabulary. Spanish and French speakers get full speech recognition in their language, and many of the destination keywords (scanner, bots, chat, arbitrage, Kelly, paper) are close enough across languages to match naturally. If a localized phrase does not catch, fall back to the English keyword — for example, say "scanner" or "whales", which are recognized regardless of UI language.

To change your interface language, go to **Settings → Language**. The next voice command uses the new language automatically; there is nothing else to configure.

## Browser Support and Requirements

Voice Commands are built on the **Web Speech API** (`SpeechRecognition`), a browser feature, not a Predite service. That has a few practical consequences worth knowing up front.

Supported browsers:

Not supported:

- **Firefox** — does not implement speech recognition, so the mic button does not appear at all (see Graceful Fallback below)

Other requirements:

## Privacy

One honest nuance about the Web Speech API: in Chrome and Edge, the *browser* may stream audio to the browser vendor's own speech service to perform the transcription, exactly as it does for any voice input on any website. That processing is governed by your browser's privacy settings and the browser vendor's policy, entirely outside Predite's control or visibility. Safari performs recognition on-device. If you prefer that no audio leaves your machine for any purpose, use Safari, or simply do not use Voice Commands and navigate by clicking — the feature is purely optional and adds nothing you cannot do with the sidebar.

Because the transcript is used only to pick a route and is never persisted, Voice Commands add no entries to your account data, your Trade Journal, or any export. For the full picture of how Predite handles data, see the Security doc.

## Graceful Fallback When Unsupported

Predite degrades cleanly when speech recognition is not available. If your browser does not implement the Web Speech API — Firefox, an older browser, or a locked-down corporate environment — **the microphone button simply does not render.** There is no broken button, no error on load, and no dead control taking up space. The rest of the dashboard works exactly as it always does, and every destination the voice feature reaches is one click away in the sidebar.

This means Voice Commands are a pure enhancement layered on top of normal navigation. Nothing in Predite *depends* on them. If they are available you get a speed boost; if not, you lose nothing.

### Troubleshooting

## Plan Availability and How It Connects to Other Features

Voice Commands themselves are **available on every plan** — Starter ($29/mo), Pro ($59/mo), and Bot ($99/mo). The feature is free navigation sugar and is not gated.

What *is* gated is where some commands take you. Voice Commands route you to a page; your plan still governs what that page lets you do:

In other words, the voice layer never bypasses plan gating — it just gets you to the door faster. If a spoken command lands you on an upgrade screen, that is the same gate you would hit by clicking.

Voice Commands pair naturally with the rest of the dashboard. The most common flow is "open scanner" to find a +EV signal, then "show whales" to confirm smart money agrees, then "go to portfolio" to size the position with the Kelly calculator. For anything beyond navigation — asking *why* a market moved, summarizing news, or reasoning about an edge — use the AI Chat widget that sits right below the mic button. Voice gets you to the room; AI Chat talks you through it.

## Related Docs