Sebastian Böck
Sebastian Böck
Right now the `num_bands` argument is used universally for all kind of filterbanks, independently if the number of bands are given as an absolute number (e.g. `MelFilterbank` or as number...
Similar to #96 (refactoring the filterbank related arguments from `FilteredSpectrogramProcessor` to `FilterbankProcessor`) it would maybe a good idea to do the same for the `LogarithmicSpectrogramProcessor` and create a new `ScalingProcessor`...
Set the filterbank and transform to MelFilterbank and dct, respectively? There's always the Cepstrogram class if other parameters are needed / wanted. Move the FFCC_\* constants into the class.
http://pytest.org/latest/fixture.html#fixtures gives some nice examples, but the tests don't have to be rewritten for `py.test` necessarily.
`FilterbankProcessor` should hold all parameters to filter a `Spectrogram` (or the like). When processing data, it should create a `Filterbank` upon being called for the first time and cache it...
Right now, madmom supports two basic processing schemes, `online` (causal) and `offline`. The former only requires past data, whereas the latter needs the whole sequence/signal to be present at once...
We should split and clarify the README to be better suited for the individual ways of distribution of the library, i.e. have separate one for PyPI, GitHub, and so on.
Unfortunately enabling `:inherited-members:` option of Sphinx globally enables them, so that all classes inheriting from numpy get tons of inherited properties and methods...
Some signatures in the online docs are broken, e.g.: - `ml.hmm` http://madmom.readthedocs.org/en/latest/modules/ml/hmm.html#madmom.ml.hmm.HiddenMarkovModel misses the parameters in the blue class line - same for `ml.nn.layers` http://madmom.readthedocs.org/en/latest/modules/ml/nn.html#madmom-ml-nn-layers - the `activate()` methods of...
https://www.python.org/dev/peps/pep-0257/ lists some conventions how dostrings should be formatted. Most of the docstrings are compatible already, some are not. Some are descriptive which violate the rule that a docstring should...