O-Engineers O-Engineers Aug 2017 | Page 42

Combining the effect of glottal pulse, vocal tract, and radiation yields a single all-pole transfer func- tion as below: Where, ‘G’ is the gain factor (Usually ignored to allow the parameterization to be independent of the signal intensity) ‘ai’ can be computed from speech signal It is generally an all-pole model for most speech sounds. By using the above transfer function we can synthesize the speech samples as a linear com- bination of the previous ‘p’ samples. Therefore, the speech production model is often called the linear prediction (LP) model, or the autoregressive model. 42