Minimum variance distortionless response spectral estimation

Wölfel, Matthias; McDonough, John

doi:10.1109/msp.2005.1511829

Cited by 74 publications

(45 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While the model order allows the overall spectral resolution to be changed, the warp factor enables the spectral resolution to be steered to lower or higher frequency regions without changing the frequency axis. Note that this is in contrast to the previously proposed warped MVDR [5,6], ACCEPTED MANUSCRIPT wherein the warp factor has an influence on both the spectral resolution and the frequency axis.…”

Section: Introductionmentioning

confidence: 59%

“…A note about the differences between the present publication and [4] is perhaps now in order. The present publication presents important background information which had been discarded in the conference publication [4] because of space limitations including: A comparison of well known and not so well known ASR front-ends on close and distant recordings in terms of word error rate and class separability.…”

Section: Introductionmentioning

confidence: 80%

“…Some years later, in 2001, MVDR was applied to speech recognition by Dharanipragada and Rao [13]. To account for the frequency resolution of the human auditory system, we have introduced warped MVDR [5,6]. It extends the MVDR Table 1 Properties of spectral estimation methods.…”

Section: Mvdr Spectral Envelopementioning

confidence: 99%

“…To achieve these objectives, we have proposed to extend the warped minimum variance distortionless response (MVDR) through a second bilinear transformation [4]. This spectral envelope estimate has two free parameters to control spectral resolution: the model order, which changes the number of linear prediction coefficients, and the warp factor.…”

Section: Introductionmentioning

confidence: 99%

“…The present publication presents important background information which had been discarded in the conference publication [4] because of space limitations including: A comparison of well known and not so well known ASR front-ends on close and distant recordings in terms of word error rate and class separability. The present publication also includes a detailed analysis and discussion of phoneme confusability.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Signal adaptive spectral envelope estimation for robust speech recognition

Wölfel

2009

Speech Communication

View full text Add to dashboard Cite

This paper describes a novel spectral envelope estimation technique which adapts to the characteristics of the observed signal. This is possible via the introduction of a second bilinear transformation into warped minimum variance distortionless response (MVDR) spectral envelope estimation. As opposed to the first bilinear transformation, however, which is applied in the time domain, the second bilinear transformation must be applied in the frequency domain. This extension enables the resolution of the spectral envelope estimate to be steered to lower or higher frequencies, while keeping the overall resolution of the estimate and the frequency axis fixed. When embedded in the feature extraction process of an automatic speech recognition system, it provides for the emphasis of the characteristics of speech features that are relevant for robust classification, while simultaneously suppressing characteristics that are irrelevant for classification. The change in resolution may be steered, for each observation window, by the normalized first autocorrelation coefficient.To evaluate the proposed adaptive spectral envelope technique, dubbed warped-twice MVDR, we use two objective functions: class separability and word error rate. Our test set consists of development and evaluation data as provided by NIST for the Rich Transcription 2005 Spring Meeting Recognition Evaluation. For both measures, we observed consistent improvements for several speaker-to-microphone distances. In average, over all distances, the proposed front-end reduces the word error rate by 4% relative compared to the widely used mel-frequency cepstral coefficients as well as perceptual linear prediction.

show abstract

Section: Introductionmentioning

confidence: 59%

Section: Introductionmentioning

confidence: 80%