How to Implement Voice Control in Node.js

Most developers reach for automatic speech recognition (ASR), also known as speech-to-text (STT) engines when they actually need Speech-to-Intent—a better alternative for enabling custom voice control in domain-specific applications. Speech-to-Intent (also called intent recognition or command recognition) extracts structured meaning from voice commands. Instead of transcribing 'Turn on the bedroom lights' to text, it returns:

intent: 'turnOn',
slots: {
  room: 'bedroom',
  device: 'lights'
}

Unlike STT engines (like Google Speech API or Whisper) that transcribe everything a user says, or natural language understanding engines (like Google Dialogflow and Amazon Lex) initially built for text-based chatbots, Rhino Speech-to-Intent directly maps spoken voice commands to actions without the overhead of full transcription.

Problem: Traditional STT sends audio to the cloud, transcribes full sentences, then parses with NLP—adding latency, privacy concerns, and complexity.

Solution: Rhino Speech-to-Intent processes commands directly on-device, extracting structured intents without cloud dependencies.

When to Use Speech-to-Intent vs. Full Transcription:

• Use Rhino when: You have predefined commands, need offline operation, require low latency, or prioritize privacy
• Use STT when: You need open-ended transcription, chatbot integration, or dictation features

This tutorial shows you how to build a voice-activated trigger system using Node.js that runs across Windows, macOS, Linux, and Raspberry Pi; perfect for smart home automation, IoT device control, industrial equipment, accessibility tools, automotive interfaces, and voice-controlled enterprise applications.

Step-by-Step: Voice Control in Node.js

Prerequisites

1. Download Node.js (v18 or newer)
2. Sign up for a Picovoice Console account and copy your AccessKey
3. Train a custom context model on the Picovoice Console and download the model file (.rhn)
4. Check that you have a working microphone or audio input

1. Install Packages

Install the Rhino Speech-to-Intent Node.js SDK and the PvRecorder Node.js SDK:

npm install @picovoice/rhino-node @picovoice/pvrecorder-node

2. Initialize the Voice Command Engine

Create an instance of Rhino, passing in your AccessKey and custom context model.

const Rhino = require("@picovoice/rhino-node");

const rhino = new Rhino(
  "${ACCESS_KEY}", // AccessKey from Picovoice Console
  "${CONTEXT_FILE_PATH}" // Your custom context model (.rhn)
);

3. Set Up Audio Capture for Intent Detection

Begin capturing audio with PvRecorder to prepare for intent detection:

const { PvRecorder } = require("@picovoice/pvrecorder-node");

// 1. Initialize and start the audio capture device
const frameLength = 512;
const recorder = new PvRecorder(frameLength);
recorder.start();

// 2. Continuously read frames of audio, which will be passed to Rhino
while (true) {
  const audioFrame = await recorder.read();
  // rhino.process(audioFrame);
}

4. Detect & Map Intents to Actions

Stream audio frames to Rhino and handle recognized inferences. When Rhino detects a possible command, it returns a RhinoInference object.

let isFinalized = false;

while (!isFinalized) {
  const audioFrame = await recorder.read();
  isFinalized = rhino.process(audioFrame);

  if (isFinalized) {
    const inference = rhino.getInference();
    if (inference.isUnderstood) {
      const intent = inference.intent;
      const slots = inference.slots;
      // take action based on inferred intent and slot values
    }
  }
}

5. Clean Up Resources

When done, stop the recorder and release resources to free memory:

recorder.stop();
recorder.release();

rhino.release();

Complete Demo: Voice Commands in Node.js

The following complete example combines all previous steps into a functional Node.js script that continuously listens for commands and logs detections to the console.

const { PvRecorder } = require("@picovoice/pvrecorder-node");
const { Rhino } = require("@picovoice/rhino-node");
const readline = require("readline");

let isRunning = true;

// Listen for spacebar
readline.emitKeypressEvents(process.stdin);
if (process.stdin.isTTY) process.stdin.setRawMode(true);

process.stdin.on("keypress", (str, key) => {
  if (key.name === "space") {
    console.log("Stopping...");
    isRunning = false;
  } else if (key.ctrl && key.name === "c") {
    isRunning = false;
  }
});

async function main() {
  let rhino = null;
  let recorder = null;

  try {
    console.log("Initializing Rhino...");
    rhino = new Rhino(
      "${ACCESS_KEY}", // AccessKey from Picovoice
      "${CONTEXT_FILE_PATH}" // Your custom context model (.rhn)
    );

    console.log("Starting intent detection... Press SPACE to stop.");
    recorder = new PvRecorder(rhino.frameLength);
    recorder.start();

    while (isRunning) {
      const audioFrame = await recorder.read();
      isFinalized = rhino.process(audioFrame);
      if (isFinalized) {
        const inference = rhino.getInference();
        if (inference.isUnderstood) {
          const intent = inference.intent;
          const slots = inference.slots;
          console.log(intent);
          console.log(slots);
        } else {
          console.log("Command not understood")
        }
      }
    }
  } catch (err) {
    console.error("Error:", err);
  } finally {
    if (recorder) {
      try {
        recorder.stop();
        recorder.release();
      } catch (e) {
        console.warn("Failed to stop/release recorder:", e);
      }
    }
    if (rhino) {
      try {
        rhino.release();
      } catch (e) {
        console.warn("Failed to release Rhino:", e);
      }
    }
    console.log("Recorder and Rhino released. Exiting.");
    process.exit(0);
  }
}

main();

This demo uses the following packages:

• @picovoice/rhino-node
• @picovoice/pvrecorder-node

For a more detailed guide, refer to the documentation:

• Rhino Speech-to-Intent Node.js Quick Start Guide
• Rhino Speech-to-Intent Node.js API Guide
• PvRecorder Node.js Quick Start Guide
• PvRecorder Node.js API Guide

For a complete demo application, check out the Rhino Speech-to-Intent Node.js Demo on GitHub.

Troubleshooting: Common Issues

Microphone Not Detected or Audio Input Fails

• Check device permissions: Ensure your app has access to the system microphone.
• Verify sampling rate: Rhino Speech-to-Intent expects 16 kHz, 16-bit mono PCM input; mismatched formats will cause errors.

No Intent Detected

• Make sure your .rhn context file matches the phrases being spoken.
• Test with clear pronunciation and limit background noise.

Enhance Your Enterprise Voice Solution

• Add Wake Word Detection: Add Porcupine Wake Word for hands-free voice commands, enabling your device to wake on "Hey [Brand Name]" without buttons.
• Integrate with Voice Activity Detection: Integrate with Cobra Voice Activity Detection to only pass audio to Rhino when someone is speaking.
• Multi-Language Support: Train custom voice command models in multiple languages to support users in different regions.