如何使IoT UWP的连续SpeechRecognition更有效？

Question

我正在尝试不断地听取IoT设备上的特定关键字，例如“Alexa”。当触发时，我想进一步识别要执行的命令。

目前我正在使用一个无限循环，因为物联网有2GB的内存并且因此滞后很多，所以效率似乎不高。虽然这个解决方案似乎适用于x64环境。

private async void SetupTheContinuousSpeechRecognitionForAlexa()
{
    try
    {
        speechRecognizerAlexa = new SpeechRecognizer();
        speechRecognizerAlexa.Timeouts.BabbleTimeout = TimeSpan.FromSeconds(0);
        speechRecognizerAlexa.Timeouts.InitialSilenceTimeout = TimeSpan.FromSeconds(0);
        speechRecognizerAlexa.Timeouts.EndSilenceTimeout = TimeSpan.FromSeconds(0);
        speechRecognizerAlexa.Constraints.Add(new SpeechRecognitionListConstraint(new string[] { "alexa", "tiger" }));
        await speechRecognizerAlexa.CompileConstraintsAsync();
        while (true)
        {
            try
            {
                var speechResult = await speechRecognizerAlexa.RecognizeAsync();
                if (speechResult.Confidence == SpeechRecognitionConfidence.High || speechResult.Confidence == SpeechRecognitionConfidence.Medium || speechResult.Confidence == SpeechRecognitionConfidence.Low)
                {
                    IsAlexaVoiceTriggered = true;
                    var helloResult = PlayTheString("How can I help you", false);
                    OnSpeechRecognitionCommandClicked(); 
                }
                else if (isSpeechEnabled)
                {
                    IsAlexaVoiceTriggered = false;
                }
            }
            catch
            {
            }
        }
    }
    catch (Exception ex)
    {
        this.log.Error("Failed in SetupTheContinousSpeechRecognitionForAlexa " + ex.Message);
    }
}

有没有其他人试图在物联网环境中实现这一目标，如果是这样，我该如何进一步优化呢？

谢谢。