search

audeering / opensmile

The Munich Open-Source Large-Scale Multimedia Feature Extractor
https://audeering.github.io/opensmile/
Other
553 stars 74 forks source link

Why the output line in csv file always have some badlines? #52

Open jazlyn-Xv opened 2 years ago

jazlyn-Xv commented 2 years ago

When I use IS09_emotion.conf to extract fixed features from a large amount of wavs and save the output in a same csv file, some of the outputs is available, because when i want to read it with pd.read_csv, it always show some bad lines with warning 'Skipping line 725: expected 386 fields, saw 737\n',and the 737 can be 738\739...I guess it has put the results of two wavs in same line,and sometimes miss some values......

chausner-audeering commented 2 years ago

If you run multiple SMILExtract instances in parallel, and they write concurrently to the same output file, the output will get corrupted as you described. You need to have each instance write in its own output file and afterwards concatenate them.