CrossSimilarityMatrix
standard mode | Music Similarity category
Inputs
- queryFeature (vector_vector_real) - input frame features of the query song (e.g., a chromagram)
- referenceFeature (vector_vector_real) - input frame features of the reference song (e.g., a chromagram)
Outputs
- csm (vector_vector_real) - 2D cross-similarity matrix of two input frame sequences (query vs reference)
Parameters
- binarize (bool ∈ {true, false}, default = false) :
- whether to binarize the euclidean cross-similarity matrix
- binarizePercentile (real ∈ [0, 1], default = 0.095) :
- maximum percent of distance values to consider as similar in each row and each column
- frameStackSize (integer ∈ [0, ∞), default = 1) :
- number of input frames to stack together and treat as a feature vector for similarity computation. Choose 'frameStackSize=1' to use the original input frames without stacking
- frameStackStride (integer ∈ [1, ∞), default = 1) :
- stride size to form a stack of frames (e.g., 'frameStackStride'=1 to use consecutive frames; 'frameStackStride'=2 for using every second frame)
Description
This algorithm computes a euclidean cross-similarity matrix of two sequences of frame features. Similarity values can be optionally binarized
The default parameters for binarizing are optimized according to [1] for cover song identification using chroma features.
The input feature arrays are vectors of frames of features in the shape (n_frames, n_features), where 'n_frames' is the number frames, 'n_features' is the number of frame features.
An exception is also thrown if either one of the input feature arrays are empty or if the output similarity matrix is empty.
References:
[1] Serra, J., Serra, X., & Andrzejak, R. G. (2009). Cross recurrence quantification for cover song identification. New Journal of Physics.