Until now, PLP_MLP feature increased the performance accuracy of
Mandarin CTS in our experiment. Now we want to try TRAPs into our system.
If I use qnstrn to train NN, the configure is as follows:
ftr1_ftr_start=${ftr1_ftr_start:=0} \
ftr1_ftr_count=${ftr1_ftr_count:=1} \
or
ftr1_ftr_start=${ftr1_ftr_start:=13} \
ftr1_ftr_count=${ftr1_ftr_count:=1} \

When I put 15 dimension TRAPs features into one file, just like PLP or
MFCC, can I set the configure as shown above to Created narrowed
stream JUST using features (0 to 1) or (13 to 14) of 15 TRAPs input
features?

If I can do this, I do not need split 15-dimension-TRAPs feature file
into each TRAPs feature files including single dimension for qnstrn
training.
Thank you.

Qingqing