Noise Reduction in Communication Microphones
Researchers at Purdue University have developed a method to remove noise from noise-corrupted speech signals. The magnitude of the noise-corrupted speech spectrum is averaged over multiple time frames to reduce the occurrence and amplitude of musical noise. During speech time frames, the algorithm determines which frequency subbands contain useful speech information and which frequency subbands contain only noise. The frequency subbands that contain only noise are subtracted off at a larger proportion, so the noise does not compete with the speech information. For the frequency subbands that contain speech, a large amount of noise is removed if the frequency subband contains relatively strong speech, and a small amount of noise is removed if the frequency subband contains relatively weak speech. This device will provide a high degree of ambient noise reduction for the talk path of communication circuits.
Superior signal processingLess musical artifacts
CircuitrySound ModulationDevicesElimination of noise-corrupted speech signals
David KozelPurdue Electrical and Computer Engineering
United States
7,209,567
USA
