Non-negative matrix deconvolution in noise robust speech recognition

Hurmalainen, Antti; Gemmeke, Jort F.; Virtanen, Tuomas

doi:10.1109/icassp.2011.5947376

Cited by 21 publications

(13 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For noise, it is not guaranteed that similar sound events will be encountered in actual use cases. In our work on AURORA-2, we saw error rates increasing by up to 60% for mismatched noises (Gemmeke et al, 8 2011b;Hurmalainen et al, 2011a). Because a noise mismatch degrades the effectiveness of speech-noise separation, and keeping a generic database for all possible noise types would be infeasible, methods for context-sensitive noise modelling are needed for practical applications.…”

Section: Pre-sampled Exemplar Basesmentioning

confidence: 84%

Modelling non-stationary noise with spectral factorisation in automatic speech recognition

Hurmalainen¹,

Gemmeke²,

Virtanen³

2013

Computer Speech & Language

Self Cite

View full text Add to dashboard Cite

Speech recognition systems intended for everyday use must be able to cope with a large variety of noise types and levels, including highly non-stationary multi-source mixtures. This study applies spectral factorisation algorithms and long temporal context for separating speech and noise from mixed signals. To adapt the system to varying environments, noise models are acquired from the context, or learnt from the mixture itself without prior information. We also propose methods for reducing the size of the bases used for speech and noise modelling by 20-40 times for better practical applicability. We evaluate the performance of the methods both as a standalone classifier and as a signal-enhancing front-end for external recognisers. For the CHiME noisy speech corpus containing non-stationary multi-source household noises at signal-to-noise ratios ranging from +9 to -6 dB, we report average keyword recognition rates up to 87.8% using a single-stream sparse classification algorithm.

show abstract

Section: Pre-sampled Exemplar Basesmentioning

confidence: 84%

Modelling non-stationary noise with spectral factorisation in automatic speech recognition

Hurmalainen¹,

Gemmeke²,

Virtanen³

2013

Computer Speech & Language

Self Cite

View full text Add to dashboard Cite

show abstract

“…The aforementioned NMF algorithms can be run on magnitude, power, and Mel-scale spectra. As an additional transformation of the spectrogram V, a sliding window can be applied as in [7], transforming the original spectrogram V to a matrix V whose columns correspond to overlapping sequences of short-time spectra in V. This provides a contextual factorization as does NMD, but with each sliding window factorized separately; whether this approach is superior to NMD seems to depend on the application [12]. Note that openBliSSART also implements inverse operations to the aforementioned transformations of the spectrogramincluding Mel filtering and sliding window-to allow proper signal reconstruction.…”

Section: Component Separation Algorithmsmentioning

confidence: 99%

“…The different DFT window sizes considered were powers of two, ranging from 2 6 to 2 12 , or 8-256 ms assuming 16 kHz sample rate. We evaluated both the RTF for both CPU and GPU computation, taking the elapsed computation time over the length of the mixed signals.…”

Section: Benchmark Performances In Supervised Speech Separationmentioning

confidence: 99%

Optimization and Parallelization of Monaural Source Separation Algorithms in the openBliSSART Toolkit

Weninger

Schuller

2012

J Sign Process Syst

View full text Add to dashboard Cite

We describe the implementation of monaural audio source separation algorithms in our toolkit openBliSSART (Blind Source Separation for Audio Recognition Tasks). To our knowledge, it provides the first freely available C++ implementation of non-negative matrix factorization (NMF) supporting the Compute Unified Device Architecture (CUDA) for fast parallel processing on graphics processing units (GPUs). Besides integrating parallel processing, openBliSSART introduces several numerical optimizations of commonly used monaural source separation algorithms that reduce both computation time and memory usage. By illustrating a variety of use-cases from audio effects in music processing to speech enhancement and feature extraction, we demonstrate the wide applicability of our application framework for a multiplicity of research and end-user applications. We evaluate the toolkit by benchmark results of the NMF algorithms and discuss the influence of their parameterization on source separation quality and real-time factor. In the result, the GPU parallelization in openBliSSART introduces double-digit speedups with respect to conventional CPU computation, enabling real-time processing on a desktop PC even for high matrix dimensions.

show abstract

“…Using exemplars in a sparse representation (SR) formulation provides significantly improved noise robustness and exemplarbased sparse representations have been successfully used for feature extraction, speech enhancement and noise robust speech recognition tasks [17][18][19][20]. These approaches model the acoustics using fixed length exemplars which are labeled at frame level and stored in the columns of a single overcomplete dictionary.…”

Section: Introductionmentioning

confidence: 99%

Data selection for noise robust exemplar matching

Yılmaz

Gemmeke

hamme

2016

2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

Exemplar-based acoustic modeling is based on labeled training segments that are compared with the unseen test utterances with respect to a dissimilarity measure. Using a larger number of accurately labeled exemplars provides better generalization thus improved recognition performance which comes with increased computation and memory requirements. We have recently developed a noise robust exemplar matching-based automatic speech recognition system which uses a large number of undercomplete dictionaries containing speech exemplars of the same length and label to recognize noisy speech. In this work, we investigate several speech exemplar selection techniques proposed for undercomplete speech dictionaries to find a trade-off between the recognition accuracy and the acoustic model size in terms of the amount of speech exemplars used for recognition. The exemplar selection criterion has be to chosen carefully as the amount of redundancy in these dictionaries is very limited compared to overcomplete dictionaries containing plenty of exemplars. The recognition accuracies obtained on the small vocabulary track of the 2 nd CHiME Challenge and the AURORA-2 database using the complete and pruned dictionaries are compared to investigate the performance of each selection criterion.

show abstract

Non-negative matrix deconvolution in noise robust speech recognition

Cited by 21 publications

References 12 publications

Modelling non-stationary noise with spectral factorisation in automatic speech recognition

Modelling non-stationary noise with spectral factorisation in automatic speech recognition

Optimization and Parallelization of Monaural Source Separation Algorithms in the openBliSSART Toolkit

Data selection for noise robust exemplar matching

Contact Info

Product

Resources

About