Underdetermined Sound Source Separation Using Power Spectrum Density Estimated by Combination of Directivity Gain

Hioka, Yusuke; Furuya, Kazuo; Kobayashi, Kazunori; Niwa, Kenta; Hori, Yoichi

doi:10.1109/tasl.2013.2248715

Cited by 51 publications

(54 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…PSD estimation in beamspace [17] is first applied to separate the coherent interfering sources. Let L (≥ K) beamformers, which focus their directivity on different angles, be applied for microphone array observation.…”

Section: Psd Estimation In Beamspacementioning

confidence: 99%

See 1 more Smart Citation

Automatic Parameter Switching of Noise Reduction for Speech Recognition

Kawase

Niwa

Hioka³

et al. 2017

Journal of Signal Processing

View full text Add to dashboard Cite

Noise reduction has been extensively studied to prevent the accuracy of automatic speech recognition (ASR) being severely degraded due to various noise sources observed in practical use. The effect of noise reduction is usually dependent on the parameters used in the noise-reduction method, which need to be tuned for different types of noise to achieve the best accuracy. We propose a method for automatically switching noise-reduction parameters in such a way that ASR accuracy is maximized. The proposed method assigns the most suitable parameter set to each speech sentence by measuring and grouping the characteristics of noise observed in each sentence. Experiments using speech sentences contaminated by various types of noise showed that the proposed method can reduce the word error rate of ASR.

show abstract

Section: Psd Estimation In Beamspacementioning

confidence: 99%

“…We previously investigated methods for estimating the PSDs of sound sources separately by looking into the temporal [16] and spatial [17,18] characteristics of each sound source. With these methods, several parameters that can only be determined empirically are needed to calculate the PSDs and coefficients of the post-filter.…”

Section: Introductionmentioning

confidence: 99%

Automatic Parameter Switching of Noise Reduction for Speech Recognition

Kawase

Niwa

Hioka³

et al. 2017

Journal of Signal Processing

View full text Add to dashboard Cite

show abstract

“…Given the sparseness and non-correlativity of the source signals, φY,Ω,τ can be modeled as Eq. (3) [9], where φΘ,Ω,τ ∈ C L and DΩ : C L → C L denote the CPSD of the sound sources inside each Θl and the gains of the BFs to the beamspace. φY,Ω,τ = DΩφΘ,Ω,τ (3) We assume Θ1 is the target beamspace.…”

Section: Xωτ = Aωmentioning

confidence: 99%

“…Applying a Wiener lter [2][3][4][5][6] to the beamforming output [7,8] is an effective way to reduce noise with several microphones. We have already proposed a PSD-estimationin-beamspace method for estimating the power spectral densities (PSDs) of the target and other sounds on the basis of the phase and amplitude differences between microphones, referred to as spatial cues [9][10][11]. The target/noise PSD estimation has been demonstrated to be robust in many circumstances, but the target PSD estimation errors sometimes drastically increase when the sound sources are not sparse, especially in very noisy environments.…”

Section: Introductionmentioning

confidence: 99%

“…Inspired by these studies, we propose a technique for integrating beamforming [9][10][11] with model-based speech enhancement [16] in a way that can adapt to every environment in real time. This technique uses xed beamforming lters (BFs) and estimates the PSD of the residual noise by using the PSD-estimation-in-beamspace method in real time.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Real-time integration of statistical model-based speech enhancement with unsupervised noise PSD estimation using microphone array

Kawase

Niwa

Fujimoto

et al. 2016

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

Self Cite

View full text Add to dashboard Cite

We propose a technique of multi-channel speech enhancement based on integration of beamforming and statistical model-based speech enhancement to clearly extract the target speech, even in very noisy environments. Conventional microphone array-based techniques estimate speech and noise power spectral densities (PSDs) from the spatial cues of the sound sources; however, their estimation errors dramatically increase when there are many noise sources. We integrated clean speech models trained in advance and the noise PSDs estimated in beamspace to compose observation models and designed a precise Wiener lter. Experiments under adverse noise conditions showed that the proposed technique signi cantly improved the signal-to-noise ratios (SNRs) compared with the conventional microphone array processing technique.Index Terms-Microphone array, beamforming, power spectral density estimation, statistical model, Wiener lter.

show abstract