private void speakButton_Click(object sender, EventArgs e)
{
string sentence = sentenceTextBox.Text;
if (sentence != "")
{
speechVisualizer.MarkerList = new List <SoundMarker>();
speechVisualizer.SetPitchPeriodSpecification(null);
string voiceName = voiceSelectionComboBox.SelectedItem.ToString();
speechSynthesizer.SetOutputToWaveFile("./tmpOutput.wav", new SpeechAudioFormatInfo(16000, AudioBitsPerSample.Sixteen, AudioChannel.Mono));
speechSynthesizer.Speak(sentence);
speechSynthesizer.SetOutputToDefaultAudioDevice();
speechSynthesizer.SelectVoice(voiceName);
speechSynthesizer.Speak(sentence);
currentSound = new WAVSound();
currentSound.LoadFromFile("./tmpOutput.wav");
double startTime = currentSound.GetFirstTimeAboveThreshold(0, 10, 20);
double endTime = currentSound.GetLastTimeAboveThreshold(0, 10, 20);
currentSound = currentSound.Extract(startTime, endTime);
speechVisualizer.SetRange(0, currentSound.GetDuration(), -32768, 32768);
speechVisualizer.SetSound(currentSound);
speechVisualizer.Invalidate();
soundTypeIdentificationButton.Enabled = true;
playSoundButton.Enabled = true;
modifySoundButton.Enabled = true;
saveSoundToolStripMenuItem.Enabled = true;
}
}
public IWRRecognitionResult RecognizeSingle(WAVSound sound)
{
// Compute the features of the current sound
sound.SubtractMean();
double startTime = sound.GetFirstTimeAboveThreshold(0, soundExtractionMovingAverageLength,
soundExtractionThreshold);
double endTime = sound.GetLastTimeAboveThreshold(0, soundExtractionMovingAverageLength,
soundExtractionThreshold);
WAVSound extractedInstance = sound.Extract(startTime, endTime);
if (extractedInstance == null)
{
return(null);
} // 20170114
extractedInstance.PreEmphasize(preEmphasisThresholdFrequency);
WAVFrameSet frameSet = new WAVFrameSet(extractedInstance, frameDuration, frameShift);
frameSet.ApplyHammingWindows(alpha);
SoundFeatureSet soundFeatureSet = new SoundFeatureSet();
List <SoundFeature> autoCorrelationFeatureList = frameSet.GetAutoCorrelationSeries("AutoCorrelation", autoCorrelationOrder);
soundFeatureSet.FeatureList.AddRange(autoCorrelationFeatureList);
List <SoundFeature> lpcAndCepstralFeatureList = frameSet.GetLPCAndCepstralSeries("LPC", lpcOrder, "Cepstral", cepstralOrder);
soundFeatureSet.FeatureList.AddRange(lpcAndCepstralFeatureList);
SoundFeature relativeNumberOfZeroCrossingsFeature = frameSet.GetRelativeNumberOfZeroCrossingsSeries("RNZC");
soundFeatureSet.FeatureList.Add(relativeNumberOfZeroCrossingsFeature);
soundFeatureSet.SetNormalizedTime();
soundFeatureSet.Interpolate(numberOfValuesPerFeature);
IWRRecognitionResult recognitionResult = new IWRRecognitionResult();
recognitionResult.SoundFeatureSet = soundFeatureSet;
if (averageSoundFeatureSetList != null)
{
foreach (SoundFeatureSet averageSoundFeatureSet in averageSoundFeatureSetList)
{
double deviation = SoundFeatureSet.GetDeviation(averageSoundFeatureSet, soundFeatureSet, weightList);
string soundName = averageSoundFeatureSet.Information;
recognitionResult.DeviationList.Add(new Tuple <string, double>(soundName, deviation));
}
recognitionResult.DeviationList.Sort((a, b) => a.Item2.CompareTo(b.Item2));
}
return(recognitionResult);
}
public WAVFrameSet(WAVSound sound, double frameDuration, double frameShift)
{
double soundDuration = sound.GetDuration();
this.frameDuration = frameDuration;
this.frameShift = frameShift;
this.frameList = new List <WAVSound>();
int numberOfFrames = (int)Math.Truncate((soundDuration - frameDuration + frameShift) / frameShift);
this.startTimeList = new List <double>();
for (int ii = 0; ii < numberOfFrames; ii++)
{
double startTime = ii * frameShift;
double endTime = startTime + frameDuration;
WAVSound frame = sound.Extract(startTime, endTime);
frameList.Add(frame);
startTimeList.Add(startTime);
}
}
private void ShowSound(WAVSound sound)
{
ClearHistory();
try // This (using try-catch) is ugly, but appears to be necessary for some hardware configurations
{
double duration = sound.GetDuration();
if (duration > viewingInterval)
{
WAVSound visibleSound = sound.Extract(duration - viewingInterval, duration);
SetSound(visibleSound);
}
else
{
SetSound(sound);
}
}
catch
{
// Nothing to do here..
}
displayBusy = false;
}
private void RunLoop()
{
Thread.Sleep(1);
DateTime utteranceStartDateTime = DateTime.Now; // Just needed for initialization.
DateTime utteranceEndDateTime = DateTime.MinValue; // Just needed for initialization.
DateTime previousUtteranceStartDateTime = DateTime.MinValue;
DateTime previousUtteranceEndDateTime = DateTime.MinValue;
DateTime recordingStartDateTime;
DateTime recordingEndDateTime;
double utteranceStartTime = 0; // In seconds, measured from the start of the current recording. (=0 just for initialization).
double utteranceEndTime;
while (running)
{
Thread.Sleep(millisecondRecordingInterval);
byte[] soundData = wavRecorder.GetAllRecordedBytes(out recordingStartDateTime, out recordingEndDateTime);
if (soundData != null)
{
if (soundData.Length > 0)
{
WAVSound sound = new WAVSound("", wavRecorder.SampleRate, wavRecorder.NumberOfChannels, wavRecorder.NumberOfBitsPerSample);
sound.AppendSamplesAsBytes(soundData);
if (showSoundStream)
{
if (!displayBusy)
{
WAVSound soundToDisplay = sound.Copy(); // 20171207: Make a new copy here, since the code below may process the sound before visualization is completed.
if (InvokeRequired)
{
this.BeginInvoke(new MethodInvoker(() => ShowSound(soundToDisplay)));
}
else
{
ShowSound(soundToDisplay);
}
}
}
// Next, remove all parts of the sound that have already been recognized, if any:
if (previousUtteranceEndDateTime > recordingStartDateTime)
{
double extractionStartTime = (previousUtteranceEndDateTime - recordingStartDateTime).TotalSeconds;
double extractionEndTime = sound.GetDuration();
sound = sound.Extract(extractionStartTime, extractionEndTime);
// Debug code, remove
/* if (sound == null) // Should not happen, unless the recognition thread is stopped using a breakpoint.
* {
*
* } */
}
if (!inUtterance)
{
utteranceStartTime = sound.GetFirstTimeAboveThreshold(0, movingAverageLength, detectionThreshold);
if (utteranceStartTime > 0)
{
double duration = sound.GetDuration();
double timeToEnd = duration - utteranceStartTime;
long ticksToEnd = TICKS_PER_SECOND * (long)(timeToEnd);
utteranceStartDateTime = recordingEndDateTime.Subtract(new TimeSpan(ticksToEnd));
if (utteranceStartDateTime > previousUtteranceEndDateTime) // True (by construction) the FIRST time.
{
inUtterance = true;
long utteranceStartTimeAsTicks = (long)(TICKS_PER_SECOND * utteranceStartTime); // Corrected 20170907 (1000000 -> 10000000)
utteranceStartDateTime = recordingStartDateTime.Add(new TimeSpan(utteranceStartTimeAsTicks));
}
}
}
else
{
double duration = sound.GetDuration();
WAVSound endOfSound = sound.Extract(duration - detectionSilenceInterval, duration);
double startTimeInEndOfSound = endOfSound.GetFirstTimeAboveThreshold(0, movingAverageLength, detectionThreshold);
if (startTimeInEndOfSound < 0) // <=> silence at the end of the sound
{
inUtterance = false;
utteranceEndDateTime = recordingEndDateTime; // recordingStartDateTime.Add(new TimeSpan(utteranceEndTimeAsTicks));
previousUtteranceStartDateTime = utteranceStartDateTime;
previousUtteranceEndDateTime = utteranceEndDateTime;
// Monitor.Enter(recognitionLockObject);
if (!recognizerBusy)
{
recognizerBusy = true;
WAVSound soundToRecognize = sound.Extract(utteranceStartTime - extractionMargin, duration).Copy();
// Monitor.Exit(recognitionLockObject);
RunRecognizer(soundToRecognize);
}
}
}
}
}
}
}