voice and peek detection

[BAT_MAN]

I am thinking to do my project in voice recognition by take peaks and valleys of each user input signal and taking average of them which i think always come unique is my approach is correct

 

[supiper]

for voice recognition you must use HMM based on features extraction (Cepstrum, LPC, RASTA-PLP etc...); you can use HTK software for that.

Added after 24 minutes:

https://htk.eng.cam.ac.uk/

 

[fal_con]

I tried this approach of analying the voice on the basis of average peak value but we can take average as unique value they vary in magnitude

 

[fam_marufi]

use simple a controller and then make a program who work on level sensor