pydrobert
is the namespace I use for the python utilities that I redistribute
open source. Subpackage XXX
can be found on PyPI with the name
pydrobert-XXX
and on Conda with the same name on the channel sdrobert
:
pip install pydrobert-XXX
conda install -c sdrobert pydrobert-XXX
pip install git+https://github.com/sdrobert/pydrobert-XXX # bleeding edge
The package can then be imported via
import pydrobert.XXX
This is student-driven code, so don't expect a stable API. I'll try to use semantic versioning, but the best way to keep functionality stable is by pinning the version in the requirements or by forking.
- pydrobert-kaldi: Pythonic
bindings for Kaldi I/O. Unlike other packages,
pydrobert-kaldi
needs no external Kaldi installation to work. Compiled for Windows, OSX, and Linux. - pydrobert-param: Intuitive
hyperparameter (de)serialization that couples with the
param
package. Includes hooks to read parameters in from a file automatically usingargparse
arguments. - pydrobert-pytorch: PyTorch modules, utilities, and various bits-and-bobs I use when building models for PyTorch. Includes gradient estimators, edit-distance-based metrics, data loaders, experiment management, attention mechanisms, and other things.
- pydrobert-speech: Digital signal processing for speech. This includes the standard Mel-scaled filter banks, and so much more. The results can be saved to NumPy, PyTorch, or Kaldi.
- pydrobert-gpyopt: Wrappers to make GPyOpt more accessible. PyPI only. Dumped in favour of Optuna.
All subpackages are Apache 2.0 licensed.
If you use any pydrobert subpackage in an academic publication, please cite as follows:
@misc{robertsonPydrobert2019,
title = {pydrobert},
url = {https://github.com/sdrobert/pydrobert},
author = {Robertson, Sean},
year = {2019}
}