ainqa-voice-assistant

Generic voice control layer for any web or Electron application. Framework-agnostic field matching via LLM. Supports Web Speech API and Whisper (DeepInfra) as STT backends.

Install

npm install ainqa-voice-assistant

Quick Start

import {
  VoiceProvider,
  VoiceAssistantBar,
  useVoice
} from 'ainqa-voice-assistant';

// 1. Wrap your app with the provider
<VoiceProvider config={{
  sttMode: 'webspeech',
  onDataExtracted: (matches) => {
    matches.forEach(m => console.log(m.fieldId, m.value));
  },
  onCommandDetected: (cmd) => {
    if (cmd === 'continue') goToNextStep();
  }
}}>
  <YourApp />
  <VoiceAssistantBar />
</VoiceProvider>

// 2. Register fields in any component
function YourForm() {
  const { registerFields } = useVoice();

  useEffect(() => {
    registerFields([
      { fieldId: 'age', label: 'Patient Age', type: 'input' },
      { fieldId: 'diabetes', label: 'Diabetes', type: 'boolean' },
      {
        fieldId: 'priority',
        label: 'Priority',
        type: 'single_select',
        options: [
          { label: 'Routine', value: 'routine' },
          { label: 'Urgent', value: 'urgent' },
        ]
      }
    ]);
  }, []);
}

STT Modes

| Mode | Privacy | Latency | Arabic | |------|---------|---------|--------| | webspeech | Google servers | ~0ms | Limited | | whisper | DeepInfra API | ~700ms | Strong |

Supported Commands

| Voice | Command fired | |-------|--------------| | "next" / "continue" / "proceed" | continue | | "stop" / "pause" | stop | | "add order" / "add lab order" | add_order | | "confirm" / "save" | confirm | | "cancel" / "discard" | cancel |