Add Custom Wake Word Detection to Your .NET C# App

Developers looking to integrate wake word detection into an enterprise .NET C# application can do so using the Porcupine Wake Word engine. Whether the goal is to enable voice activation with phrases like "Hey Jarvis," "Computer," or a custom branded wake word, Porcupine Wake Word provides a lightweight, on-device solution that delivers the same experience as Alexa, Siri, or Google Assistant—without relying on cloud services.

In this tutorial, you'll learn how to add Porcupine Wake Word into a .NET project, including package installation, custom keyword training, and implementation in code.

How to Add Wake Word Detection to a .NET App

Ensure your system meets the required .NET version:

• Windows (x86_64): .NET Framework 4.6.1+, .NET Standard 2.0+, or .NET Core 3.0+
• macOS (x86_64): .NET Standard 2.0+ or .NET Core 3.0+
• macOS (arm64), Windows (arm64), Linux (x86_64), Raspberry Pi (3, 4, 5): .NET 6.0+

1. Get Your AccessKey

Sign up for a free Picovoice Console account and obtain your AccessKey. It is required only for authentication and authorization.

2. Train Custom Wake Word Models

While Porcupine Wake Word comes with several built-in keywords, you can also create and train custom wake word models for your application using the Picovoice Console. Before training, refer to our guide on choosing an effective wake word to ensure your custom keyword works optimally within your .NET application.

3. Install the NuGet Package

Install the Porcupine Wake Word NuGet package:

dotnet add package Porcupine

4. Initialize Porcupine Wake Word

If you are using built-in keywords, use FromBuiltInKeywords. For custom keywords, use FromKeywordPaths. If you are using a non-English keyword, download the corresponding language model.

Option 1: Built-in Keywords

using Pv;

Porcupine porcupine = Porcupine.FromBuiltInKeywords(
        "${ACCESS_KEY}", // Your Picovoice AccessKey
        new List<BuiltInKeyword> {BuiltInKeyword.PORCUPINE, BuiltInKeyword.BUMBLEBEE});

Option 2: Custom Keywords

using Pv;

Porcupine porcupine = Porcupine.FromKeywordPaths(
        "${ACCESS_KEY}", // Your Picovoice AccessKey
        new List<string> {"${KEYWORD_FILE_PATH}"}, // Your custom keyword file (.ppn)
        modelPath: "${MODEL_FILE_PATH}"); // Your model file (.pv) (for non-English keywords)

5. Start Listening For Keywords

Feed audio frames from your application to Porcupine to detect keywords:

short[] GetNextAudioFrame()
{
    // ...
}

while (true)
{
    int keywordIndex = porcupine.Process(GetNextAudioFrame());
    if(keywordIndex == 0)
    {
        // detected `porcupine`
    }
    else if (keywordIndex == 1)
    {
        // detected `bumblebee`
    }
}

6. Clean Up Resources

Once you no longer need Porcupine, call Dispose() to free acquired memory:

porcupine.Dispose();

Complete Demo: Wake Word Detection in .NET

Putting it all together, here's a fully functional console demo in C#. PvRecorder is used for audio capture:

using System;
using System.Threading;
using Pv;

class PorcupineDemo
{
    static void Main()
    {
        Console.WriteLine("Initializing Porcupine & PvRecorder...");
        Porcupine porcupine = Porcupine.FromBuiltInKeywords(
                "${ACCESS_KEY}", // Your Picovoice AccessKey
                new List<BuiltInKeyword> {BuiltInKeyword.PORCUPINE, BuiltInKeyword.BUMBLEBEE});
        PvRecorder recorder = PvRecorder.Create(frameLength: porcupine.FrameLength);

        var processingThread = new Thread(() =>
        {
            while (recorder.IsRecording)
            {
                short[] audioFrame = recorder.Read();
                int keywordIndex = porcupine.Process(audioFrame);
                if(keywordIndex == 0)
                {
                    Console.WriteLine("Detected: Porcupine");
                }
                else if (keywordIndex == 1)
                {
                    Console.WriteLine("Detected: Bumblebee");
                }
            }
        });

        recorder.Start();
        processingThread.Start();

        Console.WriteLine("Listening for keyword(s)... Press Enter to stop.");
        Console.ReadLine();
        recorder.Stop();
        processingThread.Join();

        porcupine.Dispose();
        recorder.Dispose();

        Console.WriteLine("Stopped and resources released.");
    }
}

For a complete .NET application, see the Porcupine Wake Word .NET demo on GitHub.

This tutorial uses the following package:

• Porcupine Wake Word NuGet Package
• PvRecorder NuGet Package

Explore our documentation for more details:

• Porcupine Wake Word .NET Quick Start
• Porcupine Wake Word .NET API

Tips & Best Practices: Wake Word Detection

• Audio format: Ensure your audio input is recorded in the format expected by the detection engine. Porcupine Wake Word requires mono, 16-bit PCM audio. The sample rate should match porcupine.SampleRate, and the frame length should be set to porcupine.FrameLength.
• Keyword uniqueness: When selecting or creating wake words, choose keywords that are distinctive and unlikely to be confused with other common words. This improves detection accuracy. Refer to our detailed guide on choosing effective wake words.
• Efficient threading: For optimal performance, capture and process audio in a separate thread or use a non-blocking loop to prevent UI or application logic from becoming unresponsive. This ensures a smooth user experience while continuously monitoring for wake words.

Next Steps

Once wake word detection is set up, you can expand your application with additional voice functionality, such as:

• Rhino Speech-to‑Intent: handle voice commands after a wake word.
• Cheetah Streaming Speech-to-Text: transcribe speech in real time after a wake word.