Synthetic Speech Files

Speech Processing, JHU, 525.747

The examples on the class web site use a
10th order predictor, A(N+1)=30cm^2 (area past lips), 
and r0=0.9 (single zero radiation-filter coefficient):

/a/ Fs=8000 F0=70 P=40 K=55
   This means Fs=8KHz, F0=70Hz, P=40 samples, K=55 samples
   So, P/K=0.73 and P:Fs/F0=0.35

/i/ Fs=8000 F0=70 P=30 K=60
   This means Fs=8KHz, F0=70Hz, P=40 samples, K=60 samples
   So, P/K=0.5 and P:Fs/F0=0.35

To inspect the example files:

EDU>> load a10000
EDU>> sound(v,Fs)  % play vocal tract output
EDU>> sound(y,Fs)  % play vocal tract with lip radiation output
EDU>> whos         % inspect the parameters (F0, K, P, area, etc):

        Name        Size       Elements

          Fs       1 by 1             1  Sampling frequency
           N       1 by 1             1  Number of samples (8000)
       alpha       1 by 11           11  Predictor coefficients
        area       1 by 11           11  Areas
      area_A       1 by 10           10  Fant sampled areas
   area_lips       1 by 1             1  Final area (30 cm^2)
        gain       1 by 1             1  (ignore)
           k       1 by 1             1  Rosenberg K parameter
           p       1 by 1             1  Rosenberg P parameter
       pitch       1 by 1             1  F0
          r0       1 by 1             1  Radiation filter coefficient
           v       1 by 8000       8000  vocal tract output
           y       1 by 8000       8000  final output
___________________________________________________________________
Dr. Campbell, speaking for myself      The Johns Hopkins University
U.S. Dept of Defense    Digital Signal Processing: A Review Journal
j.campbell@ieee.org  http://www.apl.jhu.edu/Classes/Notes/Campbell/