Frequency-domain Adaptive Filter Research Articles

A multifilter approach to acoustic echo cancellation

Hands-free teleconferencing is increasingly frequent today. An important design consideration for any such communication tool that uses high-quality audio is the return echo caused by the acoustic coupling between the loudspeakers and microphones at each end of the conference. An echo-suppression filter (ESF) reduces the level of this return echo, increasing speech intelligibility. A new ESF has been designed based on a block frequency domain adaptive filter using the well-known least-mean-square (LMS) criteria. There are two important coefficients in LMS adaptive filters which affect how an ESF adapts to changing acoustic conditions at each end of the conference, such as double-talk conditions and moving electroacoustic transducers. Previous approaches to similar ESFs have used either a single or double pair of these coefficients, whereas the new model typically uses ten. The performance of single, double, and multifilter architectures was compared. Performance was evaluated using both empirical measurements and subjective listening tests. Speech and music were used as the stimuli for a two-way teleconferencing experiment. The new filter performed better than the single- and two-filter ESF designs, especially in conferencing conditions with frequent double talk, and the new ESF can be optimized to suit different acoustic situations.

Frequency domain adaptive filter for gravitational burst analysis

The noise of gravitational wave detectors is Gaussian but non-stationary so any good filtering technique should be adaptive. We propose a two-step adaptive procedure to identify signal and/or noise bursts, consisting of an adaptive zero phase Wiener filter in the frequency domain followed by a series of band-pass filters and an adaptive threshold algorithm to identify the events and their characteristics. This approach has been applied to simulated Virgo-like data.

A Multidelay Double-Talk Detector Combined with the MDF Adaptive Filter

The multidelay block frequency-domain (MDF) adaptive filter is an excellent candidate for both acoustic and network echo cancellation. There is a need for a very good double-talk detector (DTD) to be combined efficiently with the MDF algorithm. Recently, a DTD based on a normalized cross-correlation vector was proposed and it was shown that this DTD performs much better than the Geigel algorithm and other DTDs based on the cross-correlation coefficient. In this paper, we show how to extend the definition of a normalized cross-correlation vector in the frequency domain for the general case where the block size of the Fourier transform is smaller than the length of the adaptive filter. The resulting DTD has an MDF structure, which makes it easy to implement, and a good fit with an echo canceler based on the MDF algorithm. We also analyze resource requirements (computational complexity and memory requirement) and compare the MDF algorithm with the normalized least mean square algorithm (NLMS) from this point of view.

Improving the response of a wheel speed sensor by using frequency-domain adaptive filtering

In this paper, a frequency-domain least-mean-square adaptive filter is used to cancel noise in a wheel speed sensor embedded in a car under performance tests. In this case the relevant signal is buried in a broad-band noise background, where we have little or no prior knowledge of the signal or noise characteristics. The results of the experiments show that the signal of interest and the noise (all forms of interference, deterministic, as well as stochastic) share the same frequency band and that the filter used significantly reduced the noise corrupting the information from the sensor while it left the true signal unchanged from a practical point of view. In this paper, a signal-to-noise ratio improvement higher than 40 dB is achieved. The results of the experiment show the importance of using digital signal processing when dealing with a signal corrupted by noise.

Joint filterbanks for echo cancellation and audio coding

Joint structures for audio coding and echo cancellation are investigated, utilizing standard audio coders. Two types of audio coders are considered, coders based on cosine modulated filterbanks and coders based on the modified discrete cosine transform (MDCT). For the first coder type, two methods for combining such a coder with a subband echo canceller are proposed. The two methods are: a modified audio coder filterbank that is suitable for echo cancellation but still generates the same final decomposition as the standard audio coder filterbank, and another that converts subband signals between an audio coder filterbank and a filterbank designed for echo cancellation. For the MDCT based audio coder, a joint structure with a frequency-domain adaptive filter based echo canceller is considered. Computational complexity and transmission delay for the different coder/echo canceller combinations are presented. Convergence properties of the proposed echo canceller structures are shown using simulations with real-life recorded speech.

An approach to reducing the effect of contamination on the desired response of a frequency domain adaptive filter

In this paper, the frequency domain signal restoration problem is considered, where the desired response signal is contaminated by noise. The circumstances in which this problem arises are outlined and the impact on filter performance is described. We use a Bayesian framework to derive a solution that can be viewed as an approximate MAP estimator for the optimal filter response. The performance of this estimator is analyzed through both theory and simulation.

Iterated partitioned block frequency-domain adaptive filtering for acoustic echo cancellation

For high quality acoustic echo cancellation long echoes have to be suppressed. classical LMS-based adaptive filters are not attractive as they are suboptimal from a computational point of view. Multirate adaptive filters such as the partitioned block frequency-domain adaptive filter (PBFDAF) are good alternatives and are widely used in commercial echo cancellers nowadays. In this paper the PBFDRAP is analyzed, which combines frequency-domain adaptive filtering with so-called row action projection. Fast versions of the algorithm are derived and it is shown that the PBFDRAP outperforms the PBFDAF in a realistic echo cancellation setup.

New constraining method for partitioned block frequency-domain adaptive filters

For adaptive filters with many taps, a short processing delay, and relative low computational complexity, the partitioned block frequency-domain adaptive filter (PBFDAF) is a good choice. The computational complexity of this algorithm is significantly reduced by using the alternated constrained PBFDAF. This particular algorithm applies gradient constraints in an alternating manner. In this paper, a serious problem in performance of the alternated constrained PBFDAF is described. Furthermore, a modification is proposed that effectively alleviates this problem. By using this modification together with a gradient constraint approximation, a new efficient alternated constrained PBFDAF is developed. Compared with the fully constrained PBFDAF, this new constraining method does not show any significant performance loss and still has a much smaller computational complexity.

Stimulus artifact cancellation in the serosal recordings of gastric myoelectric activity using wavelet transform

Previous studies have shown that electrical stimulation of the stomach (i.e., gastric pacing) with appropriate parameters is a promising method for treatment of gastroparetic patients. The recording of gastric myoelectric activity (GMA) by serosal electrodes is often used to evaluate the effect of stimulation. However, the major problem with the measurement of GMA during gastric pacing is the stimulus artifacts which are often superimposed on the serosal recording and make analysis difficult. The frequency-domain adaptive filter has been used to reduce the stimulus artifacts but only with limited success. This paper describes a wavelet transform-based method for the reduction of stimulus artifacts in the serosal recordings of GMA. The key of this method lies in the use of the fuzzy set theory to select the stimulus artifact-related modulus maxima in the wavelet domain. Both quantitative and qualitative measures show that significant stimulus artifact cancellation was achieved through a series of computer simulations. Results from both single- and multichannel serosally recorded myoelectric signals during gastric pacing are presented to demonstrate the efficiency of the proposed method for the cancellation of stimulus artifacts.

Partitioned residual echo power estimation for frequency‐domain acoustic echo cancellation and postfiltering

AbstractResidual echo arises in hands‐free telephony equipment due to insufficient echo canceler convergence, but can be suppressed using a postfilter. The residual echo power spectral density is the most crucial control parameter for both frequency‐domain acoustic echo cancellation and combined residual echo and noise postfiltering. In this contribution we present and compare residual echo power spectral estimation techniques. We introduce a new partitioned block‐adaptive estimation technique delivering considerably improved residual echo estimates in strongly reverberant and noisy acoustic environments. We show that the adaptation loop of the frequency‐domain adaptive filter (FDAF) can be used simultaneously for residual echo power estimation and tracking of the echo path impulse response. In this way, the FDAF and the postfilter concept supplement each other in a true synergy with low complexity. The resulting echo and noise control system proves to be robust in double talk situations as well.

A multichannel acoustic echo canceler double‐talk detector based on a normalized cross‐correlation matrix

AbstractMultichannel acoustic echo cancellation is basically composed of two parts. One part is a multichannel system identification problem which is nontrivial to solve. The other part, which is also nontrivial, is the so‐called double‐talk detection problem. Near‐end speech detection is based on a test statistic. Recently, a new double‐talk test statistic based on the normalized cross‐correlation vector was proposed For the single‐channel case. Obviously, in the multichannel case, there are several solutions, but what should be the optimal one is not yet known. Then, a fundamental question arises: how do we deal optimally with multiple channels? In this paper, we generalize the idea of normalized cross‐correlation vector (single‐channel) to the matrix case (multichannel), derive a frequency‐domain version, and show how to combine both the multichannel frequency‐domain adaptive filter and the multichannel double‐talk detector.

周波数領域処理を用いた地上デジタル放送用逐次適応型回り込みキャンセラ

Frequency domain adaptive filters, both orthogonal transform and subband types, are introduced into loop cancellers, which were developed to cope with loop coupling and multipath distortion at relay stations in digital terrestrial broadcasting systems using OFDM (Orthogonal Frequency Division Multiplexing). An additional loop configuration and a coefficient duplicating method are also proposed and evaluated for subband-type cancellers. Computer simulation showed that frequency-domain processing has better performance than transversal-type processing.

Hybrid subband/frequency-domain adaptive systems

For many years now, subband and frequency-domain adaptive filtering techniques have been proposed for the identification of high-order FIR systems. Classical LMS-based algorithms are less attractive as their computational load is higher and their convergence behaviour for coloured inputs is worse. Subband processing has many desirable properties. However, when used to implement adaptive filters, various effects (such as residual errors and slow convergence due to aliasing) occur and reduce performance. On the other hand, it is known that frequency-domain adaptive filters do not suffer from these problems despite being (nearly) equivalent to subband adaptive filters, be it with “poor” filter banks (modulated sinc frequency responses). In this paper the operation of the frequency-domain techniques are unravelled and explained in the “subband jargon” and among other things it will be shown how aliasing is compensated for in the frequency-domain approach. This paper aims at generalising frequency-domain aliasing-compensation techniques to subband adaptive systems. Further, some design criteria for subband adaptive systems are formulated. Three realisation conditions and an alternative adaptation scheme will be proposed. Standard subband adaptive filters cannot fulfil all conditions, whereas for frequency-domain-based algorithms the design criteria are met.

A frequency-selective stepfactor control for an adaptive filter algorithm working in the frequency domain

In this contribution, a frequency-domain adaptive filter combined with a frequency-selective stepfactor control for acoustic echo cancellers is presented. The adaptive filter algorithm uses the overlap and save implementation method to calculate the error signal and the direction in which the filter impulse response should be corrected via the frequency domain with reduced numerical complexity compared to adaptive filters which use the NLMS algorithm. The frequency-selective stepfactor control improves the convergence property of the filter in environments with high background noise levels in the local room. The power spectral density functions of the excitation and the noise signal and the spectral convergence state of the compensator steer the convergence speed in different frequency bands of the adaptive filter. Therefore, the proposed stepfactor control in the frequency domain may outrange stepfactor controls which work in the time domain.

Adaptive inverse control algorithm for shock testing

An adaptive inverse control algorithm is proposed for shock testing an arbitrary specimen using an electrodynamic actuator. The purpose is to ascertain whether the specimen can survive and continue to function under severe shock conditions. The main difficulty in shock control is that the specimen dynamics vary significantly and a control algorithm is required that adapts to the characteristics of a new specimen. The control algorithm used is the adaptive inverse control method which approximates an inverse model of the loaded shaker with a finite impulse response adaptive filter, such that the reference input is reproduced at the shaker output. The standard filtered-x least mean square control structure used in the adaptive inverse control algorithm is modified to a block-processing structure, with the frequency-domain adaptive filter as the adaptation algorithm. Practical results show that the filtered-x frequency-domain adaptive filter control algorithm allows convergence of the shaker output to the assigned reference shock pulse.

Connecting partitioned frequency-domain filters in parallel or in cascade

The efficient implementation of connected filters is an important issue in signal processing. A typical example is the cascade of two filters, e.g., an adaptive filter with a time-invariant prefilter. The filtering and adaptation is carried out very efficiently in the frequency domain whenever filters with many coefficients are required. This is implemented as a block algorithm by using overlap-save or overlap-add techniques. However, in many real-time applications also, a short latency time through the system is required, which leads to a degradation of the computational efficiency. Partitioned frequency-domain adaptive filters, also known as multidelay adaptive filters, provide an efficient way for the filtering and adaptation with long filters maintaining short processing delays. This paper shows a computationally efficient way of implementing two or more partitioned frequency-domain filters in cascade or in parallel when their filter lengths are large. The methods presented require only one fast Fourier transform (FFT) and one inverse fast Fourier transform per input and output port, respectively. The FFT size can be even smaller than the length of the filters, The filters can be either time invariant or adaptive.

Lattice PFBLMS: Fast converging structure for efficient implementation of frequency-domain adaptive filters

It has been demonstrated that the unconstrained partitioned fast block LMS (PFBLMS) algorithm suffers from an inherently slow convergence rate. A solution to this problem has previously been proposed by the second author of this paper, which reduces the correlation by reducing the overlap between partitions. A drawback of this solution is that it requires more memory than the conventional PFBLMS algorithm. In this paper, we propose an alternative solution that does not require any additional memory. A lattice structure is used instead to remove the correlation and increase the convergence rate.

New Robust frequency-domain adaptive filter with colored signal input

New Robust frequency-domain adaptive filter with colored signal input

Systolic design of frequency-domain block LMS adaptive digital filters

Adaptive filtering in the frequency domain can be achieved by Fourier transformation of the input signal and independent weighting of the contents of each frequency bin. In certain applications, filtering in the frequency domain results in great improvements in convergence rate over the conventional time-domain adaptive filtering. In this paper, the use of word-level systolic arrays to implement frequency-domain adaptive filters based on the complex least mean square (LMS) algorithm is described. The transform employed is the discrete Fourier transform (DFT). The proposed architecture operates on a block-by-block basis and makes use of the parallelism inherent in the computational problem under consideration. The input and output data flow sequentially and continuously into and out of the systolic arrays at the system clock rate. During each clock period, processing elements of three different types operate in parallel. The most computationally demanding among them performs only three consecutive multiplications and two addition/subtractions per clock period thereby allowing a very high throughput and very fast block signal processing to be achieved at the expense of a delay of 2 L+1 samples between the input and the output, L being the block size.

Multidelay block Hartley domain adaptive filter

The multidelay block frequency domain (MDBFD) adaptive filter with the two-dimensional optimum block (TOB) algorithm are extended to the Hartley domain. The convergent behaviour of the MDBHD-TOB is the same as its frequency domain counterpart while computational complexity is reduced to offer simpler implementation. Because of its computational efficiency, the MDBHD-TOB adaptive filter is well suited to applications which demand the use of a long adaptive filter, as in acoustic echo cancellation.