* update: `copasul.py` restructured: all functions now included as methods in `Copasul` class. * update: `copasul_init.copa_opt_init()` * path names standardized to respective platform (so that provided minimal example should run on MS Windows without manual update of the config file) * if config input is a string (name of json file), relative paths in this file are expanded relative to its location * update: directory structure changed for packaging * as a consequence `copasul.py` cannot be run anymore on the command line. See `scripts/example_call.py` for a command line script * `copasul` now available as package which can be installed with `pip install copasul` * tests added to `tests/` folder * praat scripts and `example_call.py` moved to `scripts/` folder * documentation updated accordingly
1.2.3
* fixed: `copasul_styl.styl_polyfit()` replaced `np.int()` by `int()`, since the former is not supported anymore by numpy.
1.2.2
* changed: `copasul_init.copasul_default()`, `copasul_augment.aug_cntr_seed()`, and `copasul_clst.clst_main()`: provide `seed` for KMeans clustering and bandwidth estimation for the purpose of reproducibility
1.2.1
* added: `copasul_utils.robust_corrcoef()`
1.2.0
* general re-factoring * bugfix in copasul_export.upd_suma_feat(): IQR calculation corrected (before it was STD; affects all `*iqr` columns `summary.csv` output) * `copasul_sigproc.specmom()`: replaced `sum()` by `np.sum()` leading to slightly different spectral moment values (differences are smaller than 1e-06) * `copasul_augment.aug_prep_copy()`: keep original timestamps for automatically detected accents instead of those rounded to 2nd decimal.
1.1.0
* bugfix in copasul_styl.styl_voice_feat(): sorting pulse time stamps * `sigFunc.py` renamed to `copasul_sigproc.py` * added progress bars for stylization steps * ongoing refactoring