| 摘要 |
In a speech signal processing system for transmission or recognition, cepstrum data is normalized by subtraction from its straight-line approximation. As a result, personal and transmission characteristics are eliminated, and both data and subtraction computations are reduced.
|
