Naoto Sasaoka

A new noise reduction system based on ALE and noise reconstruction filter

A noise reduction technique is proposed to reduce wideband and sinusoidal noise in noisy speech. In real environments, the background noise includes sinusoidal noise, for example, from ventilation fans. Therefore, it is also important to reduce sinusoidal noise. The new noise reduction system uses two types of adaptive line enhancer (ALE) and a noise reconstruction filter (NRF). First, two ALEs estimate the speech components. However, since two ALEs cannot estimate unvoiced sound, the quality of the speech enhanced by the two ALEs is not enough. Thus, the NRF is used to improve the noise reduction ability. The NRF accurately estimates the background noise from the signal occupied by the noise components, which is obtained by subtracting the speech enhanced by the two ALEs from the noisy speech. The enhanced speech is obtained by subtracting the reconstructed noise from the noisy speech. Additionally, the noise reduction system with a feedback path is also proposed for more improvement of the quality of the enhanced speech.

Speech enhancement based on adaptive filter with variable step size for wideband and periodic noise

A speech enhancement system based on an adaptive line enhancer (ALE) and a noise estimation filter (NEF) has been proposed to reduce both wideband and periodic noise in noisy speech. However, it is inevitable to use a small step size which enables us to update tap coefficients of the NEF without estimating a speech signal. When a fixed step size is used, it is difficult for the speech enhancement system to reduce background noise while maintaining the high quality of enhanced speech. Thus, a variable step size is introduced to the NEF in this paper. In a speech section, the variable step size is decreased so as not to estimate speech. On the other hand, the variable step size is increased to track background noise in a non-speech section. Additionally, the variable step size is increased when the tap coefficients do not converge adequately.

Smart noise reduction system based on ALE and noise reconstruction system

The noise reduction technique to reduce wideband and sinusoidal noise in noisy speech is proposed. We have proposed the noise reconstruction system (NRS) using the linear prediction error filter (LPEF) and system identification model. However, since the sinusoidal noise has the characteristic just like the vowel, the conventional system cannot reduce the sinusoidal noise. We introduce the adaptive line enhancer (ALE) to reduce the sinusoidal noise. The ALE estimates a current signal correlated with a delayed signal. The speech signal is nonstationary signal in long time interval. On the other hand, the sinusoidal noise is stationary noise. For example, ventilating fan and engine noise. So the autocorrelation of the sinusoidal noise is maintained although that of the speech signal fades. Thus the ALE can estimate only the sinusoidal noise. Therefore, the noise reduction system based on the ALE and the NRS reduces not only the wideband noise but also the sinusoidal noise.

Noise Reduction System Based on LPEF and System Identification with Variable Step Size

We have investigated a noise reduction method based on a linear prediction error filter (LPEF) and system identification. Background noise is estimated by system identification. In the case of using a fixed step size with which tap coefficients of an adaptive filter is updated, it is difficult to reduce background noise while maintaining the high quality of enhanced speech. Therefore, a variable step size is proposed. In a speech section, a small step size is used so as not to estimate speech. On the other hand, a large step size is used to track background noise in a non-speech section.

A Noise Reduction System for Wideband and Sinusoidal Noise Based on Adaptive Line Enhancer and Inverse Filter

A noise reduction technique to reduce wideband and sinusoidal noise in a noisy speech is proposed. In an actual environment, background noise includes not only wideband noise but also sinusoidal noise, such as ventilation fan and engine noise. In this paper, we propose a new noise reduction system which uses two types of adaptive line enhancers (ALE) and a noise estimation filter (NEF). First, the two ALEs are used to estimate speech components. The first ALE is used to reduce sinusoidal noise superposed on speech and wideband noise, while the second ALE is used to reduce wideband noise superposed on speech. However, since the quality of the speech enhanced by two ALEs is not good enough due to the difficulty in estimating unvoiced sound using the two ALEs, the NEF is used to improve on noise reduction capability. The NEF accurately estimates the background noise from the signal occupied by noise components, which is obtained by subtracting the speech enhanced by two ALEs from noisy speech. The enhanced speech is obtained by subtracting the estimated noise from noisy speech. Furthermore, the noise reduction system with feedback path is proposed to improve further the quality of enhanced speech.

A study of adaptive guard interval with estimation of channel impulse response for OFDM system

In an orthogonal frequency division multiplexing (OFDM) system, a guard interval (GI) is used to remove the inter-symbol interference (ISI) due to a multipath channel. When the length of the GI is shorter than the maximum delay of the multipath channel, the ISI occurs. On the other hand, a long GI decreases transmission efficiency. In this paper, we propose an adaptive control method of the guard interval length. A channel impulse response is obtained by applying inverse fast Fourier transform (IFFT) to the estimated frequency-domain channel response. The maximum delay is then estimated from the channel impulse response, and the optimal length of the GI is decided from the maximum delay.

A study on less computational load of noise reduction method based on ALE and noise estimation filter

A noise reduction system based on adaptive line enhancer (ALE) and noise estimation filter has been proposed to reduce both wideband and sinusoidal noise in noisy speech. However, the noise reduction system uses several adaptive filters and thus the system cannot avoid increasing computation load. ALE for estimating the sinusoidal noise especially uses large numbers of taps in order to increase the estimation accuracy of the sinusoidal noise. In this paper, the noise reduction system with less computation load is proposed. The tap coefficients of the ALE have the peak by pitch period of sinusoidal noise. The proposed method takes advantage of the characteristics and decreases the number of taps.

K factor estimation for MIMO multipath channels

The channel capacity of a MIMO system depends on a Rician K factor. In the fields of a cognitive radio and link adaptation, it is required to estimate the K factor accurately. This paper proposes K factor estimation for a MIMO multipath channel. The conventional K factor estimation assumes that the MIMO channel is flat fading. Thus, the conventional method cannot estimate the K factor for MIMO multipath channel. In order to improve the estimation accuracy of a K factor, this paper adopts a multipath channel model as a MIMO Channel. In addition, since the proposed method selects high SNR paths, the estimation accuracy is more improved.

Speech enhancement using normalized cumulant-based adaptive algorithm for transient noise

Speech enhancement has been proposed to reduce noise in noisy speech. However, it is difficult to reduce the transient noise with a narrow band component which is associated with a natural oscillation of the object. In order to solve the problem, we have proposed the speech enhancement based on a linear predictor with 4th order cumulant adaptive algorithm. However, the divergence of the proposed adaptive algorithm occurs due to too large step size. Thus, this paper proposes a normalized 4th order cumulant based adaptive algorithm. The convergence of the adaptive algorithm is guaranteed by the proposed adaptive algorithm.

Active noise control with bias free pre-inverse adaptive system

Filtered-x algorithm has a possibility of making an active noise control (ANC) unstable due to the modeling error of a secondary path. A pre-inverse type ANC is proposed in order to solve the problem. The proposed ANC uses the filter which has the inverse transfer function of a secondary path before the secondary path. Whereas the filtered-x algorithm controls a primary path and a secondary path simultaneously by an adaptive filter, the pre-inverse type ANC can control a primary path and a secondary path independently. Therefore the proposed ANC is always stable. However, the adaptive filter estimating a secondary path converges on a solution with bias due to disturbance. Thus, the bias free adaptive algorithm is also proposed. The proposed adaptive algorithm takes advantage of the independence between the input signal and disturbance.