Showing papers by "Yutaka Kaneda published in 1993"

Exponentially weighted stepsize NLMS adaptive filter based on the statistics of a room impulse response

Abstract: A normalized least-mean-squares (NLMS) adaptive algorithm with double the convergence speed, at the same computational load, of the conventional NLMS for an acoustic echo canceller is proposed. This algorithm, called the ES (exponentially weighted stepsize) algorithm, uses a different stepsize (feedback constant) for each weight of an adaptive transversal filter. These stepsizes are time-invariant and weighted proportionally to the expected variation of a room impulse response. The algorithm adjusts coefficients with large errors in large steps, and coefficients with small errors in small steps. A transition formula is derived for the mean-squared coefficient error of the algorithm. The mean stepsize determines the convergence condition, the convergence speed, and the final excess mean-squared error. Modified for a practical multiple DSP structure, the algorithm requires only the same amount of computation as the conventional NLMS. The algorithm is implemented in a commercial acoustic echo canceller, and its fast convergence is demonstrated. >

Echo cancelling method and echo canceller using the same

Abstract: A received signal vector x(n), a coefficient error covariance matrix P(n) from a memory part 12 and a forgetting factor ν from a memory part 13 are provided to a gain calculating part 14 to obtain a gain vector k(n). The thus obtained gain vector k(n) and an error e(n) between an echo and an echo replica are multiplied in a multiplying part 16. The multiplied output and a filter coefficient h(n) from a memory part 18 are added together to update the latter. The thus updated filter coefficient is used as the filter coefficient of an estimated echo path (an FIR filter, for example). The coefficient error covariance matrix P(n), the gain vector k(n), the received signal vector x(n) and the forgetting factor ν are provided to an updating part 19 to update the coefficient error covariance matrix P(n), and an adjustment matrix A representing an expectation of an impulse response variation of an echo path is added to the updated coefficient error covariance matrix P(n) and the added value is used as a new coefficient error covariance matrix P(n).

Device for detecting traveling information on object

Abstract: PURPOSE: To measure the position, traveling direction, traveling speed, etc, of a traveling object at a dead angle CONSTITUTION: The position of a traveling object 13 is found in such a way that acoustic waves are radiated to a passage 24 by driving a loudspeaker 21 with a sound source signal x(k) and acoustic waves in the passage 25 are received with a microphone 22 Then an impulse response is estimated by inputting the sound receiving output y(k) of the microphone 22 and the sound source signal x(k) to an acoustic echo canceler 23 and the acoustic wave propagating time along the shortest route from the microphone 22 of the object 13 is found Finally, the position of the object 13 is found from the propagating time COPYRIGHT: (C)1995,JPO