-1

My thesis is about voice segeration. I am reading a code which has been used in this field. It most probably not needed but here is the fullCode

I can't understand something which owner code did ,

if ((fp = fopen(filename, "r")) == NULL){
    printf("Cannot open input file!\n");
    exit(0);
}

while (!feof(fp))
{
    float f;
    fscanf(fp, "%f\n", &f);
    Input[SigLength]=f;
    SigLength++;
}
fclose(fp);

float sumE=0;
for(int n=0; n<SigLength; n++)
    sumE += Input[n]*Input[n];

sumE /= float(SigLength);

for(int n=0; n<SigLength; n++)
    Input[n] *= 1000/sqrt(sumE);

return(1000/sqrt(sumE));

He opens a file which has DB of sound like 48, 200 etc..

My question why he is doing the process below ? To apply just logic is common in sound analysis ?

for(int n=0; n<SigLength; n++)
    Input[n] *= 1000/sqrt(sumE);

1 Answers1

1

This normalizes the RMS power of the data set, per unit time, but based on a measurement over the entire data set rather than the periodic or sliding type of power measurement used by a continuous AGC.

Chris Stratton
  • 33,282
  • 3
  • 43
  • 89