Segmentation - mtpearce/idyom GitHub Wiki
This app applies a peak-picking algorithm to the output of idyom to identify grouping boundaries at points of high information content (or low conditional probability). Internally, a '1' indicates that the event follows a predicted grouping boundary (e.g., the first event in a new phrase) while a '0' indicates that this is not the case.
For more information see:
Pearce, M. T., Müllensiefen, D., & Wiggins, G. (2010). The role of expectation and probabilistic learning in auditory boundary perception: A model comparison. Perception, 39, 1367-1391.
The following is an example:
CL-USER> (segmentation:idyom-segmentation 8 '(cpitch bioi deltast) '(cpitch bioi deltast) :models :ltm :threshold 1.5)
--------------------------------------
| | Positives Negatives | |
--------------------------------------
| True | 4844 67223 | 72067 |
| False | 1925 4963 | 6888 |
--------------------------------------
| | 6769 72186 | 78955 |
--------------------------------------
Precision = 0.72; Recall = 0.49; F_1 = 0.58
Mean Precision = 0.75; Mean Recall = 0.51; Mean F_1 = 0.59
NIL
CL-USER>