mbz-voice-sdk/index.js

🎙️ MBZ Voice SDK: Easily add voice recognition, Gemini-based AI replies, and TTS to any web app.

export class MBZVoiceAgent {
    constructor({ apiUrl = "/ask", lang = "en-US", speak = true } = {}) {
      this.apiUrl = apiUrl;
      this.lang = lang;
      this.speak = speak;
  
      // Setup recognition
      const Speech = window.SpeechRecognition || window.webkitSpeechRecognition;
      if (!Speech) throw new Error("SpeechRecognition not supported in this browser.");
      
      this.recognition = new Speech();
      this.recognition.lang = this.lang;
      this.recognition.interimResults = false;
  
      // Bind recognition result handler
      this.recognition.onresult = this._handleResult.bind(this);
  
      // Internal callbacks
      this._onTranscript = null;
      this._onResponse = null;
    }
  
    listen() {
      this.recognition.start();
    }
  
    mute() {
      this.speak = false;
    }
  
    unmute() {
      this.speak = true;
    }
  
    onTranscript(callback) {
      this._onTranscript = callback;
    }
  
    onResponse(callback) {
      this._onResponse = callback;
    }
  
    async _handleResult(event) {
      const text = event.results[0][0].transcript;
      if (this._onTranscript) this._onTranscript(text);
  
      try {
        const res = await fetch(this.apiUrl, {
          method: "POST",
          headers: { "Content-Type": "application/json" },
          body: JSON.stringify({ query: text }),
        });
  
        const data = await res.json();
        const reply = data.answer;
  
        if (this._onResponse) this._onResponse(reply);
  
        if (this.speak) {
          const utterance = new SpeechSynthesisUtterance(reply);
          utterance.lang = this.lang;
          utterance.pitch = 1;
          utterance.rate = 1;
          utterance.volume = 1;
  
          const voices = window.speechSynthesis.getVoices();
          const preferred = voices.find(v => v.name.includes("Google") || v.name.includes("Female"));
          if (preferred) utterance.voice = preferred;
  
          window.speechSynthesis.speak(utterance);
        }
  
      } catch (err) {
        if (this._onResponse) this._onResponse("⚠️ Failed to connect to backend.");
      }
    }
  }