hi all

I'm recognizing visual speech recognition based on htk, i have 4 image and extract 5 feature from each image. image feature extraction approach was PCA. i do this using Matlab. u could see my feature vector here:

feature vector:
frame 0:    0.8222    0.3879    0.1956    0.1484    0.0293
frame 1:    0.8165    0.4096    0.2012    0.1469    0.0312
frame 2:    1.6672    0.3984    0.1409    0.0867    0.0458
frame 3:    2.2425    0.4386    0.2018    0.0848    0.0567

after that, i make htk file by adding header and convert it to binary, i do this using Matlab too. my cpu is Intel so i save it as little endian.

after making htk binary file using matlab , i read that using hList, answer was this:

0:    8222.4343878.8591956.1751484.361 293.437
1:    8164.5084095.8052011.5661469.292 311.521
2:    16672.1703983.9101409.265 867.057 457.505
3:    22425.2094385.7412018.021 848.145 567.423

u could see that this file isn't the same as my original file,

please help me, i test different way , but why it happens,

--- (Edited on 3/27/2010 10:21 am [GMT-0500] by meslezendegi) ---