Set up classes

Issue #281 Set up classes for generic syllable recognition.

src/AnalysisConfigFiles/RecognizerConfigFiles/Ecosounds.GenericRecognizers.yml

@@ -1,5 +1,5 @@
 ---
-SpeciesName: Test Species.
+SpeciesName: GenericRecognizers
 CommonName: Generic Test
 AbbreviatedSpeciesName: TestSp
 
@@ -10,32 +10,20 @@ SegmentDuration: 60
 # SegmentOverlap: units=seconds;
 SegmentOverlap: 0
 
-#Proposed Approach:
-#1. Assume that a long recording has been broken into one-minute segments.
-#2. Convert each one-minute segment to a spectrogram.
-#3. Obtain a noise profile for each segment. This is to be used later to remove insect chorusing.
-#4. Scan the one-minute waveform and select "spike maxima" whose amplitude exceeds a decibel threshold, D.
-#5. Extract a single frame (say 512 samples) centred on each spike and convert to a spike spectrum.
-#6. Subtract the noise profile from the spike spectrum.
-#7. Smooth the remaining spectrum.
-#8. Look for evenly spaced harmonics in the smoothed spectrum.
-#    Typically the lowest harmonic will lie between 1200 Hz and 3000 Hz and the higher ones evenly spaced.
-#    This is the tricky bit due to variability but may work to use spectrum auto-correlation.
-
 # Each of these profiles will be analyzed
 Profiles:
-  #Standard: &STANDARD
-  #EventThreshold: 0.2
-  # Common settings
-  #BgNoiseThreshold: 3.0
-  Territorial:
-    # min and max of the freq band to search
-#    MinHz: 800          
-#    MaxHz: 8000
-#    MinDuration: 0.15
-#    MaxDuration: 0.8
-#    DecibelThreshold: 9.0
-  Wingbeats:
+    Territorial:
+        Algorithm: BlobRecognizer
+        Parameters:
+            # min and max of the freq band to search
+            MinHz: 800          
+            MaxHz: 8000
+            MinDuration: 0.15
+            MaxDuration: 0.8
+            DecibelThreshold: 9.0
+    Wingbeats:
+        Algorithm: OscillationRecognizer
+        Parameters:
 #    MinHz: 200          
 #    MaxHz: 2000
 #    DecibelThreshold: 6.0
@@ -60,6 +48,14 @@ Profiles:
   #  <<: *STANDARD
   #  DctDuration: 0.3
 
+
+  #Standard: &STANDARD
+  #EventThreshold: 0.2
+  # Common settings
+  #BgNoiseThreshold: 3.0
+
+
+
 # Available options (case-sensitive): [False/Never | True/Always | WhenEventsDetected]
 SaveIntermediateWavFiles: Never
 SaveIntermediateCsvFiles: false
@@ -73,5 +69,5 @@ DisplayCsvImage: false
 
 # Other config files to reference
 
-#HighResolutionIndicesConfig: "../Towsey.Acoustic.HiResIndicesForRecognisers.yml"
+HighResolutionIndicesConfig: "../Towsey.Acoustic.HiResIndicesForRecognisers.yml"
 ...

src/AnalysisPrograms/AnalysisPrograms.csproj

@@ -355,6 +355,9 @@
     <Compile Include="Recognizers\Base\RecognizerEntry.cs" />
     <Compile Include="Recognizers\Base\RecognizerResults.cs" />
     <Compile Include="Recognizers\CriniaTinnula_OBSOLETE.cs" />
+    <Compile Include="Recognizers\GenericBlobRecognizer.cs" />
+    <Compile Include="Recognizers\GenericOscillationRecognizer.cs" />
+    <Compile Include="Recognizers\GenericWhistleRecognizer.cs" />
     <Compile Include="Recognizers\IctalurusFurcatus.cs" />
     <Compile Include="Recognizers\LewiniaPectoralis.cs" />
     <Compile Include="Recognizers\LitoriaNasuta.cs" />

src/AnalysisPrograms/Recognizers/GenericBlobRecognizer.cs

@@ -0,0 +1,94 @@
+// <copyright file="GenericBlobRecognizer.cs" company="QutEcoacoustics">
+// All code in this file and all associated files are the copyright and property of the QUT Ecoacoustics Research Group (formerly MQUTeR, and formerly QUT Bioacoustics Research Group).
+// </copyright>
+
+namespace Recognizers
+{
+    using System;
+    using System.Collections.Generic;
+    using Acoustics.Shared.ConfigFile;
+    using AnalysisPrograms.Recognizers.Base;
+    using AudioAnalysisTools;
+    using AudioAnalysisTools.DSP;
+    using AudioAnalysisTools.WavTools;
+    using TowseyLibrary;
+
+    public class GenericBlobRecognizer
+    {
+        /// <summary>
+        /// THis method does the work.
+        /// </summary>
+        /// <param name="audioRecording">the recording.</param>
+        /// <param name="configuration">the config file.</param>
+        /// <param name="profileName">name of the call/event type.</param>
+        /// <param name="segmentStartOffset">where one segment is located in the total recording.</param>
+        /// <returns>a list of events.</returns>
+        public static RecognizerResults BlobRecognizer(AudioRecording audioRecording, Config configuration, string profileName, TimeSpan segmentStartOffset)
+        {
+            ConfigFile.TryGetProfile(configuration, profileName, out var profile);
+
+            // get the common properties
+            string speciesName = configuration[AnalysisKeys.SpeciesName] ?? "Pteropus species";
+            string abbreviatedSpeciesName = configuration[AnalysisKeys.AbbreviatedSpeciesName] ?? "Pteropus";
+
+            // The following parameters worked well on a ten minute recording containing 14-16 calls.
+            // Note: if you lower the dB threshold, you need to increase maxDurationSeconds
+            int minHz = profile.GetIntOrNull(AnalysisKeys.MinHz) ?? 800;
+            int maxHz = profile.GetIntOrNull(AnalysisKeys.MaxHz) ?? 8000;
+            double minDurationSeconds = profile.GetDoubleOrNull(AnalysisKeys.MinDuration) ?? 0.15;
+            double maxDurationSeconds = profile.GetDoubleOrNull(AnalysisKeys.MaxDuration) ?? 0.5;
+            double decibelThreshold = profile.GetDoubleOrNull(AnalysisKeys.DecibelThreshold) ?? 9.0;
+
+            var minTimeSpan = TimeSpan.FromSeconds(minDurationSeconds);
+            var maxTimeSpan = TimeSpan.FromSeconds(maxDurationSeconds);
+
+            //######################
+            //2. Convert each segment to a spectrogram.
+            using (AudioAnalysisTools.StandardSpectrograms.BaseSonogram sonogram = RecognizerTools.GetSonogram(configuration, audioRecording: audioRecording))
+            {
+                var decibelArray = SNR.CalculateFreqBandAvIntensity(sonogram.Data, minHz, maxHz, sonogram.NyquistFrequency);
+
+                // prepare plots
+                double intensityNormalisationMax = 3 * decibelThreshold;
+                var eventThreshold = decibelThreshold / intensityNormalisationMax;
+                var normalisedIntensityArray = DataTools.NormaliseInZeroOne(decibelArray, 0, intensityNormalisationMax);
+                var plot = new Plot(speciesName + "Blob", normalisedIntensityArray, eventThreshold);
+                var plots = new List<Plot> { plot };
+
+                //iii: CONVERT decibel SCORES TO ACOUSTIC EVENTS
+                var acousticEvents = AcousticEvent.GetEventsAroundMaxima(
+                    decibelArray,
+                    segmentStartOffset,
+                    minHz,
+                    maxHz,
+                    decibelThreshold,
+                    minTimeSpan,
+                    maxTimeSpan,
+                    sonogram.FramesPerSecond,
+                    sonogram.FBinWidth);
+
+                //iV add additional info to the acoustic events
+                acousticEvents.ForEach(ae =>
+                {
+                    ae.FileName = audioRecording.BaseName;
+                    ae.SpeciesName = speciesName;
+                    ae.Name = abbreviatedSpeciesName + profileName;
+                    ae.Profile = profileName;
+                    ae.SegmentDurationSeconds = audioRecording.Duration.TotalSeconds;
+                    ae.SegmentStartSeconds = segmentStartOffset.TotalSeconds;
+                });
+
+                acousticEvents = RecognizerTools.FilterEventsForSpectralProfile(acousticEvents, sonogram);
+
+                return new RecognizerResults()
+                {
+                    Events = acousticEvents,
+                    Hits = null,
+                    ScoreTrack = null,
+                    Plots = plots,
+                    Sonogram = sonogram,
+                };
+            }
+        }
+    }
+}

src/AnalysisPrograms/Recognizers/GenericOscillationRecognizer.cs

@@ -0,0 +1,151 @@
+// <copyright file="GenericOscillationRecognizer.cs" company="QutEcoacoustics">
+// All code in this file and all associated files are the copyright and property of the QUT Ecoacoustics Research Group (formerly MQUTeR, and formerly QUT Bioacoustics Research Group).
+// </copyright>
+
+namespace Recognizers
+{
+    using System;
+    using System.Collections.Generic;
+    using Acoustics.Shared.ConfigFile;
+    using AnalysisPrograms.Recognizers.Base;
+    using AudioAnalysisTools;
+    using AudioAnalysisTools.DSP;
+    using AudioAnalysisTools.StandardSpectrograms;
+    using AudioAnalysisTools.WavTools;
+    using TowseyLibrary;
+
+    public class GenericOscillationRecognizer
+    {
+        /// <summary>
+        /// THis method does the work.
+        /// </summary>
+        /// <param name="audioRecording">the recording.</param>
+        /// <param name="configuration">the config file.</param>
+        /// <param name="profileName">name of call/event type to be found.</param>
+        /// <param name="segmentStartOffset">where one segment is located in the total recording.</param>
+        /// <returns>a list of events.</returns>
+        public static RecognizerResults OscillationRecognizer(AudioRecording audioRecording, Config configuration, string profileName, TimeSpan segmentStartOffset)
+        {
+            ConfigFile.TryGetProfile(configuration, profileName, out var profile);
+
+            // get the common properties
+            string speciesName = configuration[AnalysisKeys.SpeciesName] ?? "Pteropus species";
+            string abbreviatedSpeciesName = configuration[AnalysisKeys.AbbreviatedSpeciesName] ?? "Pteropus";
+
+            // The following parameters worked well on a ten minute recording containing 14-16 calls.
+            // Note: if you lower the dB threshold, you need to increase maxDurationSeconds
+            int minHz = profile.GetIntOrNull(AnalysisKeys.MinHz) ?? 100;
+            int maxHz = profile.GetIntOrNull(AnalysisKeys.MaxHz) ?? 3000;
+            double minDurationSeconds = profile.GetDoubleOrNull(AnalysisKeys.MinDuration) ?? 1.0;
+            double maxDurationSeconds = profile.GetDoubleOrNull(AnalysisKeys.MaxDuration) ?? 10.0;
+            double decibelThreshold = profile.GetDoubleOrNull("DecibelThreshold") ?? 6.0;
+            double dctDuration = profile.GetDoubleOrNull("DctDuration") ?? 1.0;
+            double dctThreshold = profile.GetDoubleOrNull("DctThreshold") ?? 0.5;
+            double minOscFreq = profile.GetDoubleOrNull("MinOscilFreq") ?? 4.0;
+            double maxOscFreq = profile.GetDoubleOrNull("MaxOscilFreq") ?? 6.0;
+            double eventThreshold = profile.GetDoubleOrNull("EventThreshold") ?? 0.3;
+
+            //######################
+
+            //2. Don't use samples in this recognizer.
+            //var samples = audioRecording.WavReader.Samples;
+            //Instead, convert each segment to a spectrogram.
+            var sonogram = RecognizerTools.GetSonogram(configuration, audioRecording);
+            var decibelArray = SNR.CalculateFreqBandAvIntensity(sonogram.Data, minHz, maxHz, sonogram.NyquistFrequency);
+
+            // Call oscillation detector
+            /*
+            int scoreSmoothingWindow = 11; // sets a default that was good for Cane toad
+            Oscillations2019.Execute(
+            (SpectrogramStandard)sonogram,
+                minHz,
+                maxHz,
+                decibelThreshold,
+                dctDuration,
+                (int)Math.Floor(minOscFreq),
+                (int)Math.Floor(maxOscFreq),
+                dctThreshold,
+                eventThreshold,
+                minDurationSeconds,
+                maxDurationSeconds,
+                scoreSmoothingWindow,
+                out var scores,
+                out var acousticEvents,
+                //out var hits,
+                segmentStartOffset);
+*/
+            Oscillations2012.Execute(
+                (SpectrogramStandard)sonogram,
+                minHz,
+                maxHz,
+                //decibelThreshold,
+                dctDuration,
+                (int)Math.Floor(minOscFreq),
+                (int)Math.Floor(maxOscFreq),
+                dctThreshold,
+                eventThreshold,
+                minDurationSeconds,
+                maxDurationSeconds,
+                out var scores,
+                out var acousticEvents,
+                out var hits,
+                segmentStartOffset);
+
+            // prepare plots
+            double intensityNormalisationMax = 3 * decibelThreshold;
+            var normThreshold = decibelThreshold / intensityNormalisationMax;
+            var normalisedIntensityArray = DataTools.NormaliseInZeroOne(decibelArray, 0, intensityNormalisationMax);
+            var plot1 = new Plot(speciesName + " Wing-beat band", normalisedIntensityArray, normThreshold);
+            var plot2 = new Plot(speciesName + " Wing-beat Osc Score", scores, eventThreshold);
+            var plots = new List<Plot> { plot1, plot2 };
+
+            // ######################################################################
+
+            // add additional information about the recording and sonogram properties from which the event is derived.
+            acousticEvents.ForEach(ae =>
+            {
+                ae.FileName = audioRecording.BaseName;
+                ae.SpeciesName = speciesName;
+                ae.Name = abbreviatedSpeciesName + profileName;
+                ae.Profile = profileName;
+                ae.SegmentDurationSeconds = audioRecording.Duration.TotalSeconds;
+                ae.SegmentStartSeconds = segmentStartOffset.TotalSeconds;
+                var frameOffset = sonogram.FrameStep;
+                var frameDuration = sonogram.FrameDuration;
+                ae.SetTimeAndFreqScales(frameOffset, frameDuration, sonogram.FBinWidth);
+
+                //UNCOMMENT following lines to get spectral profiles of the Wingbeat events.
+                /*    double[,] spectrogramData = sonogram.Data;
+                    int maxBin = (int)Math.Round(8000 / sonogram.FBinWidth);
+                    double startSecond = ae.EventStartSeconds - ae.SegmentStartSeconds;
+                    int startFrame = (int)Math.Round(startSecond / sonogram.FrameStep);
+                    int frameLength = (int)Math.Round(ae.EventDurationSeconds / sonogram.FrameStep);
+                    int endFrame = startFrame + frameLength;
+
+                    // get only the frames from centre of the acoustic event
+                    var subMatrix = DataTools.Submatrix(spectrogramData, startFrame + 10, 0, endFrame - 10, maxBin);
+                    var spectrum = MatrixTools.GetColumnAverages(subMatrix);
+                    var normalisedSpectrum = DataTools.normalise(spectrum);
+                    normalisedSpectrum = DataTools.filterMovingAverageOdd(normalisedSpectrum, 11);
+                    var maxId = DataTools.GetMaxIndex(normalisedSpectrum);
+                    var hzMax = (int)Math.Ceiling(maxId * sonogram.FBinWidth);
+                    string name = "BeatSpectrum " + (ae.SegmentStartSeconds / 60) + "m" + (int)Math.Floor(startSecond) + "s hzMax" + hzMax;
+                    var bmp2 = GraphsAndCharts.DrawGraph(name, normalisedSpectrum, 100);
+
+                    //Set required path
+                    bmp2.Save(Path.Combine(@"C:\PATH", name + ".png"));
+                    */
+            });
+
+            return new RecognizerResults()
+            {
+                Events = acousticEvents,
+                Hits = null,
+                ScoreTrack = null,
+                Plots = plots,
+                Sonogram = sonogram,
+            };
+        }
+
+    }
+}