Class HighLevelFeaturePipelines
- java.lang.Object
-
- com.tagtraum.audiokern.audioprocessor.HighLevelFeaturePipelines
-
public final class HighLevelFeaturePipelines extends Object
HighLevelFeaturePipelines.- Author:
- Hendrik Schreiber
-
-
Method Summary
All Methods Static Methods Concrete Methods Modifier and Type Method Description static SignalPipeline<AudioBuffer,Float>
createAverageNoveltyPipeline(String id, int windowSize, int hopsize, DistanceFunction<float[]> distanceFunction, int maxFramesToProcess)
Calculates the arithmetic mean of the buffer returned by acreateNoveltyPipeline(String, int, int, com.tagtraum.jipes.math.DistanceFunction, int)
pipeline.static SignalPipeline<AudioBuffer,AudioBuffer>
createNoveltyPipeline(String id, int windowSize, int hopsize, DistanceFunction<float[]> distanceFunction, int maxFramesToProcess)
Transforms to mono, normalizes to -1/+1, applies the given window, applies a Hamming window, then transforms using FFT andstatic SignalPipeline<AudioBuffer,Float>
createNoveltyStandardDeviationPipeline(String id, int windowSize, int hopsize, DistanceFunction<float[]> distanceFunction, int maxFramesToProcess)
Calculates the standard deviation of the buffer returned by acreateNoveltyPipeline(String, int, int, com.tagtraum.jipes.math.DistanceFunction, int)
pipeline.
-
-
-
Method Detail
-
createNoveltyPipeline
public static SignalPipeline<AudioBuffer,AudioBuffer> createNoveltyPipeline(String id, int windowSize, int hopsize, DistanceFunction<float[]> distanceFunction, int maxFramesToProcess)
Transforms to mono, normalizes to -1/+1, applies the given window, applies a Hamming window, then transforms using FFT and- Parameters:
id
- id to collect any results underwindowSize
- window size in frames, recommended is 4096 for audio sampled at 44.1kHz (~93ms) and a hopsize of 2048hopsize
- hop size, recommended is half the window sizemaxFramesToProcess
- maximum number of frames to processdistanceFunction
- distance function to compute the distance between two FFT spectra- Returns:
- a novelty pipeline
- See Also:
DistanceFunctions.createCosineDistanceFunction(int, int)
,Novelty
-
createAverageNoveltyPipeline
public static SignalPipeline<AudioBuffer,Float> createAverageNoveltyPipeline(String id, int windowSize, int hopsize, DistanceFunction<float[]> distanceFunction, int maxFramesToProcess)
Calculates the arithmetic mean of the buffer returned by acreateNoveltyPipeline(String, int, int, com.tagtraum.jipes.math.DistanceFunction, int)
pipeline.- Parameters:
id
- idwindowSize
- window sizehopsize
- hopsizedistanceFunction
- distance function to compute the distance between two FFT spectramaxFramesToProcess
- max frames to process- Returns:
- pipeline that computes average novelty
- See Also:
Novelty
-
createNoveltyStandardDeviationPipeline
public static SignalPipeline<AudioBuffer,Float> createNoveltyStandardDeviationPipeline(String id, int windowSize, int hopsize, DistanceFunction<float[]> distanceFunction, int maxFramesToProcess)
Calculates the standard deviation of the buffer returned by acreateNoveltyPipeline(String, int, int, com.tagtraum.jipes.math.DistanceFunction, int)
pipeline.- Parameters:
id
- idwindowSize
- window sizehopsize
- hopsizedistanceFunction
- distance function to compute the distance between two FFT spectramaxFramesToProcess
- max frames to process- Returns:
- pipeline that computes novelty standard deviation
- See Also:
Novelty
-
-