Here is the output from running the create-dataset.js script
Total labels => count: 980, length: 3.8 hours, average length: 13.86
Outliers labels => count: 930, length: 3.6 hours, average length: 13.88
Used labels => count: 767, length: 2.7 hours, average length: 12.88
characters=" '(),-.:;?abcdeghijklmnoprstuvxyñāīūḍḷṃṅṇṭ"
characters=" '(),-.:;?xංඅආඉඊඋඌඑඔකඛගඝඞචඡජඣඤටඨඩඪණතථදධනපඵබභමයරලවසහළ්ාිීුූෙො"
speakers={"wdevananda":214,"oshadir":213,"obhasa":149,"lankananda":191}
Total labels => count: 63610, length: 480.0 hours, average length: 27.17
Usable labels => count: 20584, length: 46.9 hours, average length: 8.21
Used labels => count: 8990, length: 20.0 hours, average length: 8.01
{
paragraph: 3357,
centered: 739,
heading: 2423,
gatha: 2429,
unindented: 42
}
characters=" '(),-.:;?abcdeghijklmnoprstuvxyñāīūḍḷṃṅṇṭ"
characters=" '(),-.:;?xංඅආඉඊඋඌඑඔකඛගඝඞචඡජඣඤටඨඩඪණතථදධනපඵබභමයරලවසහළ්ාිීුූෙො"
Extracted audio from flac files in 57.90 seconds
In the releases you can find the tar archive with all the wav files. You can download it and extract it as follows
cat pali_dataset.tar.bz2.part* > pali_dataset.tar.bz2
tar -xjf pali_dataset.tar.bz2