/HDA_speechCommand

Deep Learning Project for speech recognition

Primary LanguageJupyter Notebook

HDA_speechCommand

Deep Learning Project for speech recognition The project has been developed with IPython using Jupyter Notebook and has two main notebooks:

  • Dataset and Feature Dumper contains the routine for downloading the dataset, process it and save the used variable for Processing notebook
  • Processing is the plain notebook with data selection

In the dependencies folder there are two .py files:

  • functions.py where there are all the function used in the code
  • models.py where there are all the model used in the code

In the CompiledNotebook folder is possible to find pre-run notebook with confusion matrices.

More code is in:

  • OnlineProcessing is a simple proposal for online processing
  • ExtraCode contains training methods for networks, a multiOutput inception, the hyperparamer optimization and a possible weighted categorical loss