Create spec database
-
align annotation naming scheme in AnnotationTableReader with other methods in data_handling module
-
improve handling of wave files with different duration (allow user to specify a standard length, files that are shorter will be padded; files that are longer will be split)
-
replicate folder structure inside database file (so data can be easily be split up into train, validation, test)
-
write spectrograms to database file one at a time, query for file size to get more accurate size determination
Q: The name 'orig_file' seems a little clumsy compared to the others ('start', 'end', 'label'). Could we simply call it 'filename'? SOLVED
I think this is ready to merge!