DSP - Task7 - LPC analysis
ESTIMATION OF PARAMETERS OF AR-MODEL, LPC SPECTRAL ANALYSIS
- AR model of the signal
- Evaluate autoregressive coefficients of the analyzed signal using function lpc (parameters of AR model). Set the order of AR model to p=16 and evaluate it for signal length N = 512
- Use short part of voice speech vm0.bin for evaluation purposes
- Compare short-time spectra using DFT and LPC approach
- Smoothed estimation of spectrum using LPC is given as frequency responce of sysnthesis filter of AR-model ( function freqz )
- Draw both estimation into one figure !!
- Analyse the influence of model gain G to the results
( Gain G should be evaluated as the second output parameter of function lpc )
- Analyse changes in the estimation of smoothed spectrum with respect ot changing order of AR model, i.e. p=16, 10, 6. Explain it !!
- Example: Compare the spectral resolution for DFT a LPC estimations.
- Choose suitable order for AR model !
- Evaluate spectral estimation for both cases (LPC and DFT) for
N=64 and N= 256.
- Use Hamming window weighting.
- sinusoidal: f=13 Hz, fs= 100 Hz
- 2 sinusoidals of close frequences: f1=13 Hz, f2=15 Hz, fs= 100 Hz