Heratio Help Center article. Category: Viewers & Media.

Voice Commands

Overview

Heratio includes a voice command system for hands-free navigation, search, dictation, and AI-powered image description. Powered by the Web Speech API (browser-native).

Getting Started

1. Click the microphone button in the navbar or the floating button (bottom-right)
2. Say a command (e.g. "browse", "search for photographs", "help")
3. The system responds with spoken feedback and on-screen toasts

Right-click the mic button to open a text input for typing commands manually.

Command Categories

Navigation

• "go home" / "browse" / "go to admin" / "go to settings"
• "go to donors" / "go to accessions" / "go to repositories"
• "browse archive" / "browse library" / "browse museum" / "browse gallery"
• "go back" / "next page" / "previous page"
• "search for [term]"

Record Reading

• "read metadata" — read all populated fields aloud
• "read title" / "read description"
• "describe image" — AI description via LLaVA (images only)
• "read PDF" — read PDF transcript text
• "what type of file" — report file type

AI Image Description

• "describe image" / "AI describe" — generate description
• "save to description" / "save to alt text" / "save to both"
• "discard" — discard generated description

Media Detection

• PDFs: offers to read OCR/transcript text if available
• Videos/audio: offers to read transcript if available
• Non-OCRd PDFs: notifies user text is not readable

Dictation

• "start dictating" — dictate into focused text field
• "stop dictating" — return to command mode
• Punctuation: "period", "comma", "question mark", "new line", etc.

Voice Control

• "disable voice" / "voice off" — disable until re-enabled
• "enable voice" / "voice on" — re-enable
• "keep listening" — continuous mode
• "stop listening" — single command mode

Accessibility

• "where am I" — announce current page
• "how many results" — announce result count
• "help" — show command list

Settings

Admin > AHG Settings > Voice & AI:

• Language (11 languages)
• Confidence threshold
• Continuous listening mode
• Floating button visibility
• Hover-read (TTS on hover)
• Speech rate
• LLM provider (local Ollama / cloud Anthropic / hybrid)

Browser Support

Voice commands require HTTPS and microphone permission.