BASS stands for Basic Audio System Solutions. It consists of a set of utilities functions that comes in handy when doing signal processing.
BASS was designed with the objective of being a utility librairy for Machine Learning preprocessing.
For now, the main functionnalities of BASS are mainly centered around Automatic Speech Recognition.
In the future, BASS might get extended to a lot of different Natural Language Processing tasks.
The mellib.py file contains methods to compute a Mel Frequency Cepstral Coefficients transform.
Its use is very simple. The user can refer to the documentation of the function for more customization.
from mellib import mfcc
mfcc = mfcc(signal)The augmentation.py file contains a class that allows an easy way to augment MFCC.
It follows the implementation described in the specaugment article published by the Google Brain team [1].
Its use is again, very simple. The user can refer to the documentation of the function for more customization.
from augmentation import Specaugment
augmented_mfcc = Specaugment.augment(mfcc)[1] Park, D. S., Chan, W., Zhang, Y., Chiu, C. C., Zoph, B., Cubuk, E. D., & Le, Q. V. (2019). Specaugment: A simple data augmentation method for automatic speech recognition.