OpenBCI data processing for subvocal recognition

• This is the repo to turn OpenBCI raw data into spectrograms and to train the model on said data.

1_openbci_to_wavfile.ipynb

• Checks the latest recordings in the SavedData folder from OpenBCI, takes the array of values of each channel
• and turns it into a wavfile.

2_wav_to_spectrogram.ipynb

• Before executing this code you first need to split the data using the process_audio.py from subvocalization_stream,
• once you split the wavfiles, this code transforms the individual wavfiles into sepctrograms.

3_spectrogram_to_csv.ipynb

• This script will get the file location of each spectrogram and list them into a csv file for training the model.

4_subvocal_train_model.ipynb

• This script will train the model using the assigned csv file.

5_subvocal_model_find_outliers.ipynb

• Using the trained model we can apply it to the dataset to find outliers and create a csv file to exclude said outlier data.