Realtime Phoneme Classfication with MFCC

Code base for project "Build a human-in-the-loop psychophysics speech synthesis simulator for a brain-computer interface to restore speech"

How to run

• all configurations are done by editing config/default.yaml file, and here are some configuration entries that you need to edit before running
  • root_path: this should be the absolute path to parent folder of data & dict folder
  • run_name: the checkpoint and report will be saved / loaded with name like model_${run_name}.pt

Prerequisites

• pytorch
• torchaudio
• scikit-learn
• hydra
• tqdm
• loguru