The speech-shaped and babble noise files that were used to create the dataset are available here:
Link for zip file