[Subject Prev][Subject Next][Thread Prev][Thread Next][Date Index][Thread Index]

[hts-users:02004] Problem of 1-order data stream in training and generation

Hi all,


I have encountered a strange problem while using hts-2.1. The problem is: when I set a separated data stream with order=1, windim=3 (window file like the standard demo script) in training data, I get a wrong generate parameter value at the beginning and end 3 or more  frames. The other frames seem correct. For example

The training data is like:

 -13.0472       -10.5036       -10.6337       -10.5734       -10.7509       -10.6383 ……-10.5473       -10.5321       -10.6249       -10.8142       -10.4404       -10.6346

The generate data is:

-1.4591       -5.6805       -8.5919       -11.9597       -13.4174       -18.5253………-12.8879       -14.6535       -14.5206       -9.3618       -6.4378       -1.1382

I use window.pl in demo script to add window. The delta and delta-delta computation for the beginning 3 frames is like

-13.047  -5.252  15.591

-10.504   1.207  -2.674  

-10.634  -0.035   0.190 


I cannot fix the problem. Whether the algorithm or the code or the data are not appropriate. Has anyone tried to train and generate 1-order data stream ever before?


Thanks very much!

Best regards!

Xi Wang (汪曦)

Beijing University of Posts and Telecommunications
Pattern Recognition & Intelligent Systems Lab

[hts-users:02013] Re: Problem of 1-order data stream in training and generation, Keiichiro Oura