One of my study is about Single Channel Noise Suppressor. This method can effectively suppress a noise from a noisy speech by using the proposed variable speech PDF (Probability Density Function). ( More )
This research topic is also about single chanel noise suppression.We utilize a lattice filter with low computation load to suppress a noise from a noisy speech.
We can easily implement this method on a DSP (Digital Signal Processor). Especially, this method is useful to suppress a noise from a low SNR noisy speech. ( More )
The noise suppressors mentiond above have been developed to reduce a stationary noise. The zero phase domain method reduces both of stationary and impulsive noises with the same procedure. ( More )
We have been studying a howling canceller based on distance measurement. The howling canceller is effective to reduce howling in a seminor room where a loud speaker is fixed on a wall or a cell. Measurment results of the distance between the loudspeaker and the microphone gives frequency candidates of howling. We adaptively reduce them. ( More )
We are investigating an adaptive notch filter which can remove a sinusoidal signal whose frequency is unknown. Convergence speed and estimation accuracy of the adaptive notch filter are the most important factors. We have proposed a parallel notch filter, where with the same estimation accuracy, it converges faster than the conventional adaptive notch filter. ( More )
Implementing a general comb filter is very easy, because it has
a simple structure. But, its frequency response is very constraint.
We are investigating a design method of a comb filter whose notch
frequencies, gains, and bandwidths are adjustable. ( More )
Hearable devices have been developing as one of wearable devices. To add effective and convenient functions to hearable devices, we are trying to implement technologies we developed. i More j
