• Title/Summary/Keyword: Variational model composition

Search Result 3, Processing Time 0.019 seconds

PCA-based Variational Model Composition Method for Roust Speech Recognition with Time-Varying Background Noise (시변 잡음에 강인한 음성 인식을 위한 PCA 기반의 Variational 모델 생성 기법)

  • Kim, Wooil
    • Journal of the Korea Institute of Information and Communication Engineering
    • /
    • v.17 no.12
    • /
    • pp.2793-2799
    • /
    • 2013
  • This paper proposes an effective feature compensation method to improve speech recognition performance in time-varying background noise condition. The proposed method employs principal component analysis to improve the variational model composition method. The proposed method is employed to generate multiple environmental models for the PCGMM-based feature compensation scheme. Experimental results prove that the proposed scheme is more effective at improving speech recognition accuracy in various SNR conditions of background music, compared to the conventional front-end methods. It shows 12.14% of average relative improvement in WER compared to the previous variational model composition method.

Speech enhancement method based on feature compensation gain for effective speech recognition in noisy environments (잡음 환경에 효과적인 음성인식을 위한 특징 보상 이득 기반의 음성 향상 기법)

  • Bae, Ara;Kim, Wooil
    • The Journal of the Acoustical Society of Korea
    • /
    • v.38 no.1
    • /
    • pp.51-55
    • /
    • 2019
  • This paper proposes a speech enhancement method utilizing the feature compensation gain for robust speech recognition performances in noisy environments. In this paper we propose a speech enhancement method utilizing the feature compensation gain which is obtained from the PCGMM (Parallel Combined Gaussian Mixture Model)-based feature compensation method employing variational model composition. The experimental results show that the proposed method significantly outperforms the conventional front-end algorithms and our previous research over various background noise types and SNR (Signal to Noise Ratio) conditions in mismatched ASR (Automatic Speech Recognition) system condition. The computation complexity is significantly reduced by employing the noise model selection technique with maintaining the speech recognition performance at a similar level.

A size-dependent quasi-3D model for wave dispersion analysis of FG nanoplates

  • Karami, Behrouz;Janghorban, Maziar;Shahsavari, Davood;Tounsi, Abdelouahed
    • Steel and Composite Structures
    • /
    • v.28 no.1
    • /
    • pp.99-110
    • /
    • 2018
  • In this paper, a new size-dependent quasi-3D plate theory is presented for wave dispersion analysis of functionally graded nanoplates while resting on an elastic foundation and under the hygrothermaal environment. This quasi-3D plate theory considers both thickness stretching influences and shear deformation with the variations of displacements in the thickness direction as a parabolic function. Moreover, the stress-free boundary conditions on both sides of the plate are satisfied without using a shear correction factor. This theory includes five independent unknowns with results in only five governing equations. Size effects are obtained via a higher-order nonlocal strain gradient theory of elasticity. A variational approach is adopted to owning the governing equations employing Hamilton's principle. Solving analytically via Fourier series, these equations gives wave frequencies and phase velocities as a function of wave numbers. The validity of the present results is examined by comparing them with those of the known data in the literature. Parametric studies are conducted for material composition, size dependency, two parametric elastic foundation, temperature and moisture differences, and wave number. Some conclusions are drawn from the parametric studies with respect to the wave characteristics.