摘要 |
<p>An apparatus for encoding speech includes circuits (2,9) which sample a speech signal and generate digital data representative of the signal. An encoder (10,11) connected to the sampling circuits encodes the digital data by linear predictive coding and generates parameters representing the speech signal. The parameters are transformed (12) to obtain sum and difference polynomials. The roots of the sum and difference polynomials are determined (13) producing a series of line spectrum data. A storage device (14,15) stores the line spectrum data. In one example, the line spectrum data are quantised non-uniformly in the frequency domain and the roots of the polynomials are found by evaluating those polynomials at the quantisation frequencies only. <IMAGE></p> |