A5080537604- Speech and Audio Processing
- Blind Source Separation Techniques
- Advanced Adaptive Filtering Techniques
- Music and Audio Processing
- Microfluidic and Bio-sensing Technologies
- Electrostatics and Colloid Interactions
- Microfluidic and Capillary Electrophoresis Applications
- Statistical Methods and Inference
- Orbital Angular Momentum in Optics
- Bayesian Methods and Mixture Models
- Acoustic Wave Phenomena Research
The University of Tokyo
2017-2019
Nanzan University
2019
In previous studies, acoustical levitation in the far-field was limited to particles. Here, this paper proposes "boundary hologram method," a numerical design technique generate static and stable field for macroscopic non-spherical rigid bodies larger than sound wavelength λ. This employs boundary element formulation approximate acoustic radiation force torque applied body by discretizing surface, which is an explicit function of transducer's phase amplitude. Then, drive phased array...
In this paper, we propose a new framework called independent deeply learned matrix analysis (IDLMA), which unifies deep neural network (DNN) and independence-based multichannel audio source separation. IDLMA utilizes both pretrained DNN models statistical independence between sources for the separation, where time-frequency structures of each are iteratively optimized by while enhancing estimation accuracy spatial demixing filters. As generative model, introduce complex heavy-tailed...
In this paper, statistical-model generalizations of independent low-rank matrix analysis (ILRMA) are proposed for achieving high-quality blind source separation (BSS). BSS is a crucial problem in realizing many audio applications, where the sources must be separated using only observed mixture signal. Many algorithms solving have been proposed, especially history component and nonnegative factorization. particular, ILRMA can achieve highest performance music or speech mixtures, assumes both...
In this paper, we address a multichannel audio source separation task and propose new efficient method called independent deeply learned matrix analysis (IDLMA). IDLMA estimates the demixing in blind manner updates time-frequency structures of each using pretrained deep neural network (DNN). Also, introduce complex Student's t-distribution as generalized generative model including both Gaussian Cauchy distributions. Experiments are conducted music signals with training dataset, results show...
Independent low-rank matrix analysis (ILRMA) is a fast and stable method of blind audio source separation. Conventional ILRMAs assume time-variant (super-)Gaussian models, which can only represent signals that follow super-Gaussian distribution. In this article, we focus on ILRMA based generalized Gaussian distribution (GGD-ILRMA) propose new type GGD-ILRMA adopts sub-Gaussian for the model. We update scheme called iterative projection homogeneous models (GIP-HSM) obtain...
In this paper, we generalize a source generative model in state-of-the-art blind separation (BSS), independent low-rank matrix analysis (ILRMA). ILRMA is unified method of frequency-domain component and nonnegative factorization can provide better performance for audio BSS tasks. To further improve the stability separation, introduce an isotropic complex Student's t-distribution as model, which includes Gaussian distribution used conventional ILRMA. Experiments are conducted using both music...
Independent low-rank matrix analysis (ILRMA) is a fast and stable method for blind audio source separation. Conventional ILRMAs assume time-variant (super-)Gaussian models, which can only represent signals that follow super-Gaussian distribution. In this paper, we focus on ILRMA based generalized Gaussian distribution (GGD-ILRMA) propose new type of GGD-ILRMA adopts sub-Gaussian the model. By using update scheme called iterative projection homogeneous obtain convergence-guaranteed rule...
In this letter, we propose a new blind source separation method, independent low-rank matrix analysis based on generalized Kullback-Leibler divergence. This method assumes time-frequency-varying complex Poisson distribution as the generative model, which yields convex optimization in spectrogram estimation. The experimental evaluation confirms proposed method's efficacy.
In this paper, we generalize a source generative model in state-of-the-art blind separation (BSS), independent low-rank matrix analysis (ILRMA). ILRMA is unified method of frequency-domain component and nonnegative factorization can provide better performance for audio BSS tasks. To further improve the stability separation, introduce an isotropic complex Student's $t$-distribution as model, which includes Gaussian distribution used conventional ILRMA. Experiments are conducted using both...
In this paper, we address a multichannel audio source separation task and propose new efficient method called independent deeply learned matrix analysis (IDLMA). IDLMA estimates the demixing in blind manner updates time-frequency structures of each using pretrained deep neural network (DNN). Also, introduce complex Student's t-distribution as generalized generative model including both Gaussian Cauchy distributions. Experiments are conducted music signals with training dataset, results show...
Independent low-rank matrix analysis (ILRMA) is a fast and stable method for blind audio source separation. Conventional ILRMAs assume time-variant (super-)Gaussian models, which can only represent signals that follow super-Gaussian distribution. In this paper, we focus on ILRMA based generalized Gaussian distribution (GGD-ILRMA) propose new type of GGD-ILRMA adopts sub-Gaussian the model. By using update scheme called iterative projection homogeneous obtain convergence-guaranteed rule...