摘要 |
The invention relates to a signal processing method and system, as well as a linear phase filter. The system comprises an input means 1 receiving a signal having a component with a relatively slow variation and a component with a relatively fast variation which contains phoneme information, means permitting the said signal to be distributed over a certain frequency range, means 3, 4 for sampling the said signal between one extremity and the other of the frequency range of the signal, then between the other extremity and the first, and so on, and means 5 for elimination of the component with relatively long variation, the relatively rapid component thus extracted being delivered to an output terminal 6. The invention applies to voice recognition. <IMAGE> |