| 摘要 |
A front-end audio processing system comprises: a signal separation unit, used for performing separation processing on an acquired signal to obtain an effective signal and a reference signal; a first processing unit, used for analyzing the effective signal to remove a low-frequency noise signal from the effective signal; a second processing unit, used for respectively receiving the reference signal and the effective signal on which the low-frequency noise signal removal has been performed, and removing a self-noise signal from the effective signal to form a pure audio signal by using a preset algorithm according to the reference signal; a comparison unit, used for comparing the pure audio signal with a preset threshold; and a calculation unit, used for enlarging the pure audio signal in the state in which the pure audio signal is less than the preset threshold, and decreasing the pure audio signal in the state in which the pure audio signal is not less than the preset threshold. The system fills the gap of an embedded operating system that is commonly seen in the current market in the aspect of front-end processing in the field of intelligent robots. The frame can provide the function of front-end noise reduction for back-end speech recognition application on the basis of not modifying code of the existing embedded operating system, and has relatively high system expandability and flexibility. |
