| 摘要 |
The method involves generating at least a sound datum from a sound signal (x) indicating the degree of existence of a predetermined characteristic of the sound signal. The sound signal is classified to vocabulary or to noise dependent on the sound datum. The predetermined characteristic is a measure for the excitation of a determined sound signal frequency, which is stronger for speech signals than for noise. Preferably, the energy content of the sound signal is determined in a predetermined frequency range as sound datum containing possibly the determined sound signal frequency.
|
