ENH: Add vocalypy.segment module #40
Description
- As discussed in discussion: segmenting algorithms #5
- I think we want to have an explicit module
vocalpy.segmentthat further contains methods grouped by the schema of Kemp et al 2000: energy-based methods, metric-based methods, and model-based methods - Energy-based methods can be further split into
audioandspectrogram- audio will be: amplitude threshold + audio method from Koe
-
endpoint_detection, AKA Rabiner-Sambur algorithm - whatever evsonganaly uses -- is there a name?
amplitude_threshold - Harma
-
- spectrogram will be:
- Lasseck
- vocalseg -- "dynamic thresholding"
- ava, that also uses spectrogram
- usvseg: ENH: Add Python implementation of usvseg to
vocalpy.signal.segment#43
- audio will be: amplitude threshold + audio method from Koe
- We will want a
segment.metricsmodule too; this should start out withprecisionandrecallwith atolerance/collarparameter