/// <summary>
/// This method returns foward (spectral peak) tracks enclosed in spectral events.
/// It averages dB log values incorrectly but it is faster than doing many log conversions.
/// </summary>
/// <param name="sonogram">The spectrogram to be searched.</param>
/// <returns>A list of acoustic events containing foward tracks.</returns>
public static (List <EventCommon> Events, double[] CombinedIntensity) GetForwardTracks(
SpectrogramStandard sonogram,
ForwardTrackParameters parameters,
TimeSpan segmentStartOffset)
{
var sonogramData = sonogram.Data;
int frameCount = sonogramData.GetLength(0);
int binCount = sonogramData.GetLength(1);
int nyquist = sonogram.NyquistFrequency;
double binWidth = nyquist / (double)binCount;
int minBin = (int)Math.Round(parameters.MinHertz.Value / binWidth);
int maxBin = (int)Math.Round(parameters.MaxHertz.Value / binWidth);
double minDuration = parameters.MinDuration.Value;
double maxDuration = parameters.MaxDuration.Value;
double decibelThreshold = parameters.DecibelThreshold.Value;
var converter = new UnitConverters(
segmentStartOffset: segmentStartOffset.TotalSeconds,
sampleRate: sonogram.SampleRate,
frameSize: sonogram.Configuration.WindowSize,
frameOverlap: sonogram.Configuration.WindowOverlap);
//Find all spectral peaks and place in peaks matrix
var peaks = new double[frameCount, binCount];
for (int row = 0; row < frameCount; row++)
{
for (int col = minBin + 1; col < maxBin - 1; col++)
{
if (sonogramData[row, col] < decibelThreshold)
{
continue;
}
// if given matrix element is greater than in freq bin either side
bool isPeak = (sonogramData[row, col] > sonogramData[row, col - 1]) && (sonogramData[row, col] > sonogramData[row, col + 1]);
if (isPeak)
{
peaks[row, col] = sonogramData[row, col];
}
}
}
var tracks = GetForwardTracks(peaks, minDuration, maxDuration, decibelThreshold, converter);
// initialise tracks as events and get the combined intensity array.
// list of accumulated acoustic events
var events = new List <SpectralEvent>();
var combinedIntensityArray = new double[frameCount];
// The following lines are used only for debug purposes.
//var options = new EventRenderingOptions(new UnitConverters(segmentStartOffset.TotalSeconds, sonogram.Duration.TotalSeconds, nyquist, frameCount, binCount));
//var spectrogram = sonogram.GetImage(doHighlightSubband: false, add1KHzLines: true, doMelScale: false);
// Initialise events with tracks.
foreach (var track in tracks)
{
//Following line used only for debug purposes. Can save as image.
//spectrogram.Mutate(x => track.Draw(x, options));
var maxScore = decibelThreshold * 5;
var scoreRange = new Interval <double>(0, maxScore);
var ae = new ChirpEvent(track, scoreRange)
{
SegmentStartSeconds = segmentStartOffset.TotalSeconds,
SegmentDurationSeconds = frameCount * converter.SecondsPerFrameStep,
Name = "noName",
};
events.Add(ae);
// fill the intensity array
var startRow = converter.FrameFromStartTime(track.StartTimeSeconds);
var amplitudeTrack = track.GetAmplitudeOverTimeFrames();
for (int i = 0; i < amplitudeTrack.Length; i++)
{
combinedIntensityArray[startRow + i] = Math.Max(combinedIntensityArray[startRow + i], amplitudeTrack[i]);
}
}
List <EventCommon> returnEvents = events.Cast <EventCommon>().ToList();
// Combine coincident events that are stacked one above other.
// This will help in some cases to combine related events.
if (parameters.CombinePossibleHarmonics)
{
returnEvents = CompositeEvent.CombinePotentialStackedTracks(events, parameters.HarmonicsStartDifference, parameters.HarmonicsHertzGap);
}
// Combine events that are temporally close and in the same frequency band.
// This will help in some cases to combine related events.
if (parameters.CombinePossibleSyllableSequence)
{
var timeDiff = TimeSpan.FromSeconds(parameters.SyllableStartDifference);
returnEvents = CompositeEvent.CombineSimilarProximalEvents(events, timeDiff, parameters.SyllableHertzGap);
}
return(returnEvents, combinedIntensityArray);
}
public void TestForwardTrackAlgorithm()
{
// Set up the recognizer parameters.
var parameters = new ForwardTrackParameters()
{
MinHertz = 500,
MaxHertz = 6000,
MinDuration = 0.2,
MaxDuration = 1.1,
DecibelThreshold = 2.0,
CombinePossibleHarmonics = false,
HarmonicsStartDifference = TimeSpan.FromSeconds(0.2),
HarmonicsHertzGap = 200,
CombinePossibleSyllableSequence = false,
SyllableStartDifference = 0.2,
SyllableHertzGap = 300,
};
//Set up the virtual recording.
int samplerate = 22050;
double signalDuration = 13.0; //seconds
// set up the config for a virtual spectrogram.
var sonoConfig = new SonogramConfig()
{
WindowSize = 512,
WindowStep = 512,
WindowOverlap = 0.0, // this must be set
WindowFunction = WindowFunctions.HANNING.ToString(),
NoiseReductionType = NoiseReductionType.Standard,
NoiseReductionParameter = 0.0,
Duration = TimeSpan.FromSeconds(signalDuration),
SampleRate = samplerate,
};
var spectrogram = this.CreateArtificialSpectrogramToTestTracksAndHarmonics(sonoConfig);
//var image1 = SpectrogramTools.GetSonogramPlusCharts(spectrogram, null, null, null);
//results.Sonogram.GetImage().Save(this.outputDirectory + "\\debug.png");
var segmentStartOffset = TimeSpan.Zero;
var plots = new List <Plot>();
var(spectralEvents, dBArray) = ForwardTrackAlgorithm.GetForwardTracks(
spectrogram,
parameters,
segmentStartOffset);
// draw a plot of max decibels in each frame
double decibelNormalizationMax = 5 * parameters.DecibelThreshold.Value;
var dBThreshold = parameters.DecibelThreshold.Value / decibelNormalizationMax;
var normalisedDecibelArray = DataTools.NormaliseInZeroOne(dBArray, 0, decibelNormalizationMax);
var plot1 = new Plot("decibel max", normalisedDecibelArray, dBThreshold);
plots.Add(plot1);
var allResults = new RecognizerResults()
{
NewEvents = new List <EventCommon>(),
Hits = null,
ScoreTrack = null,
Plots = new List <Plot>(),
Sonogram = null,
};
// combine the results i.e. add the events list of call events.
allResults.NewEvents.AddRange(spectralEvents);
allResults.Plots.AddRange(plots);
// effectively keeps only the *last* sonogram produced
allResults.Sonogram = spectrogram;
// DEBUG PURPOSES COMMENT NEXT LINE
this.SaveTestOutput(
outputDirectory => GenericRecognizer.SaveDebugSpectrogram(allResults, null, outputDirectory, "ForwardTrack"));
Assert.AreEqual(23, allResults.NewEvents.Count);
var @event = (SpectralEvent)allResults.NewEvents[4];
Assert.AreEqual(2.0, @event.EventStartSeconds, 0.1);
Assert.AreEqual(2.5, @event.EventEndSeconds, 0.1);
Assert.AreEqual(1720, @event.LowFrequencyHertz);
Assert.AreEqual(2107, @event.HighFrequencyHertz);
@event = (SpectralEvent)allResults.NewEvents[11];
Assert.AreEqual(6.0, @event.EventStartSeconds, 0.1);
Assert.AreEqual(6.5, @event.EventEndSeconds, 0.1);
Assert.AreEqual(2150, @event.LowFrequencyHertz);
Assert.AreEqual(2580, @event.HighFrequencyHertz);
}
