[Kaldi-users] Feature extraction arguments for online recognizers

[Mailing l. u. f. U. C. a. U. <kal...@li...>]

Hi,


I wanted to try online-audio-server-decode-faster with some *.mdl files (trained using non-default FE params) I have. There don't seem to be any feature extraction arguments (window length, number of FFT bins). How should I go about this task?


Thanks!