Characterization of Non-Stationary Channels Using Mismatched Wiener Filtering

Ispas, Adrian; Dörpinghaus, Meik; Ascheid, Gerd; Zemen, Thomas

doi:10.1109/tsp.2012.2223688

Cited by 17 publications

(17 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They can be obtained as sub-matrices of the channel autocorrelation matrix R h = E hh H by extracting Δ t -spaced rows and/or columns. See [17] for further details.…”

Section: A Time-variant Frequency-flat Channelmentioning

confidence: 99%

On the Optimum Number of Hypotheses for Adaptive Reduced-Rank Subspace Selection

Hofer

Zemen

2015

2015 IEEE 82nd Vehicular Technology Conference (VTC2015-Fall)

Self Cite

View full text Add to dashboard Cite

Intelligent transport systems (ITS) require lowlatency dependable wireless communication links inbetween vehicles as well as between vehicles and the infrastructure. In vehicular communication scenarios communication channels are time-and frequency (doubly) dispersive and the channel statistics are non-stationary, i.e., they change over time. Hence, the design of appropriate channel estimators is challenging. Recently an adaptive reduced-rank channel estimation technique for nonstationary time-variant channel estimation was introduced by Zemen and Molisch, 2012. This technique uses a hypothesis test to obtain an estimate of the current channel statistics on a per frame basis. The optimum number of hypotheses is not known. In this paper we present new empirical insights on the optimum choice of the number of hypotheses for the hypothesis test for non-stationary time-variant channel estimation. With these considerations the complexity of an adaptive reduced-rank channel estimator can be reduced and its performance improved.

show abstract

“…They can be obtained as sub-matrices of the channel autocorrelation matrix R h = E hh H by extracting Δ t -spaced rows and/or columns. See [17] for further details.…”

Section: A Time-variant Frequency-flat Channelmentioning

confidence: 99%

On the Optimum Number of Hypotheses for Adaptive Reduced-Rank Subspace Selection

Hofer

Zemen

2015

2015 IEEE 82nd Vehicular Technology Conference (VTC2015-Fall)

Self Cite

View full text Add to dashboard Cite

show abstract

“…Note here that the acoustic models were trained with non-dysarthric clean speech utterances, while they were tested with noisy dysarthric speech utterances obtained by artificially adding a babble noise and an office noise with SNRs of 10 and 15 dB. Table 2 compares the average word error rates (WERs) of a baseline ASR system and three ASR systems employing a conventional Wiener filter [7], the CV-dependent Wiener filter [11], and the proposed noise reduction method. As shown in the table, an ASR system using the proposed noise reduction method provided the lowest WERs for both dysarthric groups.…”

Section: Performance Evaluationmentioning

confidence: 99%

“…Thus, their performance was limited when the ASR system was deployed in real-world applications. To alleviate the effect of background noise on dysarthric ASR, various kinds of noise reduction methods, including spectral subtraction [5], minimum mean square error log-spectral amplitude (MMSE-LSA) [6], and Wiener filtering [7], have been applied to noisy dysarthric speech. However, since these noise reduction methods have been developed for non-dysarthric speech rather than for dysarthric speech in noisy environments, they do not reflect the various characteristics of dysarthric speech [8].…”

Section: Introductionmentioning

confidence: 99%

Reducing Speech Noise for Patients with Dysarthria in Noisy Environments

Seong

Park

Kim

2014

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYDysarthric speech results from damage to the central nervous system involving the articulator, which can mainly be characterized by poor articulation due to irregular sub-glottal pressure, loudness bursts, phoneme elongation, and unexpected pauses during utterances. Since dysarthric speakers have physical disabilities due to the impairment of their nervous system, they cannot easily control electronic devices. For this reason, automatic speech recognition (ASR) can be a convenient interface for dysarthric speakers to control electronic devices. However, the performance of dysarthric ASR severely degrades when there is background noise. Thus, in this paper, we propose a noise reduction method that improves the performance of dysarthric ASR. The proposed method selectively applies either a Wiener filtering algorithm or a Kalman filtering algorithm according to the result of voiced or unvoiced classification. Then, the performance of the proposed method is compared to a conventional Wiener filtering method in terms of ASR accuracy.

show abstract

“…system and restrict ourselves to linear processing but drop the assumption of stationarity for a quasi-stationary behaviour, whereby the space-time covariance matrix can be assumed to be stationary -and therefore only dependent on the lag parameter τ -for sufficiently short time windows [9].…”

Section: Data Model and Predictionmentioning

confidence: 99%

“…We will carry forward n since it is well known that the wind signal is non-stationary and develop an approximately stationary solution in Section 2.2. Calculating R ee [n] using (7) yields a quadratic expression in W n , Equation (9).…”

Section: Mmse Predictormentioning

confidence: 99%