Is my process correct in Speech recognition using DWT(wavlet) ?

I planned to use DWT for feature extraction in speech recognition. My steps are as follows.

1. Speech Signal -Wav file of 16kHz

2. Pre Processing -VAD -De-noising -Pre -emphasis -Dynamic time warping 3. Frame Blocking & Windowing 4. Feature Extraction -Discrete Wavelet Transform 5. Recognition Module

The doubt is i have to use DTW. It says by definition that we can find an optimal alignment between two given (time-dependent) sequences. But i need to align the time for N number of files. So that we can recognition. please help me in that.