Henning Puder

Step-size control for acoustic echo cancellation filter—an overview

Abstract In this paper we present an overview about several approaches for controlling the step size for adaptive echo cancellation filters in hands-free telephones. First an optimal step size is derived. For the determination of this step size the power of a non-measurable signal has to be estimated. Detection and estimation methods for the determination of this power and for the determination of the optimal step size can be grouped into four classes. For each class, several principles, which differ in their reliability and in their complexity, are presented. Possibilities for combining elements from each class to create an entire step size control unit are also described. An outlook on detector combinations based on neuronal networks concludes this paper.

Adaptive feedback cancellation for realistic hearing aid applications

Acoustic feedback is a well-known phenomenon in hearing aids. Under certain conditions it causes the so-called howling effect, which is highly annoying for the hearing aid user and limits the maximum amplification of the hearing aid. The standard adaptive feedback cancellation algorithms suffer from a biased adaptation if the input signal is spectrally colored, as it is for speech and music signals. Due to this bias distortion artifacts (entrainment) are generated. In this paper, we present a sub-band feedback cancellation system which combines decorrelation methods with a new realization of a known non-parametric variable step size. To apply this step size in the context of adaptive feedback cancellation, a method to estimate the signal power of the desired input signal, i.e., without feedback, is necessary. A major part of this paper is spent with the theoretical derivation of this estimate. Furthermore, the complete system is evaluated extensively for several speech and music signals as well as for different feedback scenarios in simulations with feedback paths measured in concrete applications as well as for real-time simulations with hearing aid dummies. Both use hearing loss compensation methods as applied in physical hearing aids. The performance is measured in terms of being able to prevent entrainment and to react to feedback path changes. For both simulation setups the system shows a good performance with respect to the two performance measures. Furthermore, the overall feedback cancellation method relies only on few parameters, shows a low computational complexity, and therefore has a strong practical relevance.

Kalman-filters in subbands for noise reduction with enhanced pitch-adaptive speech model estimation

In this paper, we present an approach for a Kalman filter which enhances single channel speech signals disturbed by car noise. For the Kalman filters, the speech and noise signals are modelled by AR processes and the corresponding models have to be estimated, based on the noisy speech. To avoid high model orders, the filtering is performed in subbands. We especially focus on the estimation of the speech model: First, standard model estimation methods are investigated and the method giving the best stable model is then utilized as a preliminary solution. To enhance the models without generating musical tones, a new method is proposed that utilizes estimates of the speakers pitch frequency. The algorithm we present has less than 39 ms delay and thus fulfills the ETSI requirements for mobile telephony [1].

Implementation of a hands-free car phone with echo cancellation and noise-dependent loss control

An implementation of a complete hands-free telephone system on a low-cost DSP with 16-bit fixed-point arithmetic is presented. The prototype, consisting of an adaptive echo canceller, was developed for an industrial customer. The main emphasis is laid on the control unit of the system being necessary for a reliable performance in all environments and the attenuation introduced by the loss control unit. Keeping the loss as small as possible, which is most important for a natural communication including full double-talk capabilities, requires exact knowledge of the current system performance and the background noise level. In case of background noise the applied total loss is adjusted to the power of the noise. The measures we used and their application to our system control are discussed.

A TF distribution for disturbed and undisturbed speech signals and its application to noise reduction

Abstract All quadratic time–frequency (TF) representations that provide the TF-shift covariance property belong to the well-known Cohen class. In recent years, several publications have dealt with the problem of choosing the kernel function with respect to the signal to be analyzed. This paper investigates a solution for disturbed and undisturbed speech signals aimed at achieving a good TF analysis for the noise reduction problem. First, we introduce a speech model that defines a certain class of signals. Based on the speech model we develop a new compound kernel consisting of the Margenau–Hill and the Smoothed Pseudo-Wigner distribution and demonstrate its superior performance compared with some popular distributions (Zhang–Sato distribution, Pseudo-Wigner distribution, Smoothed Pseudo-Wigner distribution, S-method). Finally, the so-called ‘Smoothed Margenau–Hill distribution’ is used to design a time–frequency filter for the noise reduction algorithm based on the time-variant Wiener filter for which we obtain promising results.

Decorrelation measures for stabilizing adaptive feedback cancellation in hearing aids

In this contribution we describe an adaptive feed-back cancellation (FBC) system realized with 48 sub-band filters. As core procedure we propose a combination of two decorrelation measures to stabilize and optimally control the adaptation. We show that especially this combination of pre-whitening and frequency shift allows realizing three major steps for a fast and reliable FBC in real hearing aids. First, the adaptation bias is removed. Second, an optimal adaptation control can be realized, and third, we show that a differentiation between feedback and tonal input signals is possible. The latter can be used for an additional improvement of the adaptation control.