Hosting for preprocessed data sets in the Na language, for use in NLP. The original data are online in the "Na" repository on GitHub and the audio is online in the Pangloss Collection (instructions for batch download are available here).
Preprocessed version 0.9 of the data set (in the folder named v0dot9_Jan2018) was created by Oliver Adams. It was hosted, at one point, here (simply note the change of the directory name from <feat> to <wav>).
These preprocessed sets were produced by means of Oliver Adams's script na.py.