This is my entry for the TensorFlow Speech Recognition Challenge.
Files are numbered in order I created them, not necessarily the order to run them. Final pipeline:
05_run_opensmile.py
to run OpenSmile voice activity detection for all the files06_max_vad.py
to extract the max from each OpenSmile outputGCommandsPytorch/02_train_all_no_crossvalidate.py
to train neural network on all the training dataGCommandsPytorch/04_predict_with_prob.py
to predict using neural network03b_postprocess_unknown_threshold.py
to combine VAD and CNN output to create final predictions