search

audeering / opensmile

The Munich Open-Source Large-Scale Multimedia Feature Extractor
https://audeering.github.io/opensmile/
Other
569 stars 75 forks source link

Prosody extractor output binary output.csv on Linux #23

Closed stasbel closed 2 years ago

stasbel commented 3 years ago
  1. install opensmile for Linux from source
  2. run ./build/progsrc/smilextract/SMILExtract -C config/prosody/prosodyShs.conf -I my.wav -O prosody.csv
  3. prosody.csv appears to be binary file and is unreadable by text editors
stasbel commented 3 years ago

with opensmile-python, I can't manage to get it work and specify my hop_length: there is just no options for that

stasbel commented 3 years ago

@frankenjoe

stasbel commented 3 years ago

I want to somehow extract F0 and loudness with your library and my specific hop length (for alignment along stft len).

chausner-audeering commented 3 years ago

prosody.csv appears to be binary file and is unreadable by text editors

The problem is that you specified -O which in this case is used to output in HTK format. If you take a look at the end of prosodyShs.conf, you see it includes shared/standard_data_output_lldonly.conf.inc. In this file, the output command-line parameters are defined. You can see that -O is for HTK output:

filename=\cm[output(O){output.htk}:output HTK binary file for LLD. Use ? as value to disable]

For CSV output, you'll need to use -csvoutput instead:

filename=\cm[csvoutput{?}:output csv file for LLD, disabled by default ?, only written if filename given]

chausner-audeering commented 2 years ago

I hope your question has been answered, thus I'll close this. If not, feel free to reopen the issue.