Speech recognition in a reverberant environment using matched filter array (MFA) processing and linguistic-tree maximum likelihood linear regression (LT-MLLR) adaptation

Raghavan, Prabhu; Renomeron, Richard J.; Che, ChiWei; Yuk, Dong-Suk; Flanagan, James L.

doi:10.1109/icassp.1999.759785

Cited by 4 publications

(2 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Because the feature extraction is performed on the output of a filter-and-sum operation, the sequence of cepstral vectors can be expressed as a function of the filter coefficients of all microphone filters h i [k]. If we concatenate the parameters of all filters into a supervector h and define y j (h) as the vector of the observations for frame j expressed as a function of these filter parameters h, then the vector of cepstral coefficients for frame j can be expressed as (2) where represents the Mel-frequency cepstral vector for frame j and represents the matrix of the weighting coefficients of the Mel filters.…”

Section: Speech Recognizer-based Microphone Array Processingmentioning

confidence: 99%

See 1 more Smart Citation

Speech recognizer-based microphone array processing for robust hands-free speech recognition

Seltzer

Raj

Stern

2002

IEEE International Conference on Acoustics Speech and Signal Processing

View full text Add to dashboard Cite

We present a new array processing algorithm for microphone array speech recognition. Conventionally, the goal of array processing is to take distorted signals captured by the array and generate a cleaner output waveform. However, speech recognition systems operate on a set of features derived from the waveform, rather than the waveform itself. The goal of an array processor used in conjunction with a recognition system is to generate a waveform which produces a set of recognition features which maximize the likelihood for the words that are spoken, rather than to minimize the waveform distortion. We propose a new array processing algorithm which maximizes the likelihood of the recognition features. This is accomplished through the use of a new objective function which utilizes information from the recognition system itself, obtained in an unsupervised manner, to optimize the parameters of a filter-and-sum array processor. Using the proposed method, improvements in word error rate of up to 36% over conventional methods are achieved on real microphone array tasks in a wide range of environments.

show abstract

Section: Speech Recognizer-based Microphone Array Processingmentioning

confidence: 99%

“…Many microphone array processing algorithms proposed over the years have been able to achieve a substantial improvement in the quality of the output speech signal, and these methods have been used as front-ends to speech recognition systems (e.g. [1] [2]). …”

Section: Introductionmentioning

confidence: 99%