Digital processing of speech signals

Multi-appliance recognition system with hybrid SVM/GMM classifier in ubiquitous smart home

In this paper we explain a fully automatic system for airplane detection and tracking based on wavelet transform and Support Vector Machine (SVM). By using 50 airplane images in different situations, models are developed to recognize airplane in the first frame of a video sequence. To train a SVM classifier for classifying pixels belong to objects and background pixels, vectors of features are built. The learned model can be used to detect the airplane in the original video and in the novel images. For original video, the system can be considered as a generalized tracker and for novel images it can be interpreted as method for learning models for object recognition. After airplane detection in the first frame, the feature vectors of this frame are used to train the SVM classifier. For new video frame, SVM is applied to test the pixels and form a confidence map. The 4th level of Daubechies's wavelet coefficients corresponding to input image are used as features. Conducting simulations, it is demonstrated that airplane detection and tracking based on wavelet transform and SVM classification result in acceptable and efficient performance. The experimental results agree with the theoretical results.

Airplane detection and tracking using wavelet features and SVM classifier

Nonlinear subspace clustering using curvature constrained distances

In many speech separation and enhancement techniques, establishing a statistical model like a Vector Quantization (VQ) is a must to handle the so-called model-based approaches. It is also desirable to establish a trade-off between sparsity and accuracy in the quantizer. To do so, in this paper we present split-VQ for sinusoidal parameters. We observed that sinusoidal parameters including amplitudes and frequencies, are most capable to be used as our features for split-VQ since they can be easily mapped to a tree-like structure. We demonstrate that using such split-VQ along with fixed dimension sparse sinusoidal parameters can significantly result in better source model compared with common STFT feature vectors in terms of objective and subjective measures in model-based approaches like monaural sound separation.

Model-based monaural sound separation by split-VQ of sinusoidal parameters

In pattern recognition, support vector machines (SVM) as a discriminative classifier and Gaussian mixture model as a generative model classifier are two most popular techniques. Current state-of-the-art systems try to combine them together for achieving more power of classification and improving the performance of the recognition systems. Most of recent works focus on probabilistic SVM/GMM hybrid methods but this paper presents a novel method for SVM/GMM hybrid pattern classification based on training data selection. This system uses the output of the Gaussian mixture model to choose training data for SVM classifier. Results on databases are provided to demonstrate the effectiveness of this system. We are able to achieve better error-rates that are better than the current systems.

Pattern Classiffication using SVM with GMM Data Selection Training Methode

The lack of a flexible analysis model has been introduced as an important issue in different applications like source separation In this paper, a fixed dimension modified sinusoid model (FD-MSM) is proposed for analysis of all audible signals consisting of speech, music and their mixtures Employing the peak picking in Meldomain gives rise to a fixed number of parameters in the proposed FDMSM, which is desired in clustering algorithms like VQ (vector quantization) or GMM (gaussian mixture model), commonly used for source separation scenarios Applying the proposed FD-MSM to various audible signals, it is observed that the resulting signal is perceptually indistinguishable from the original

A Fixed Dimension Modified Sinusoid Model (FD-MSM) for Single Microphone Sound Separation

High power amplification of linear modulation schemes which exhibit fluctuating envelopes, invariably leads to the generation of distortion and intermodulation products. In this paper, an adaptive pre-distortion technique is introduced to counter-balance the AM/AM and AM/PM nonlinear effects of the transmit power amplifier. The proposed adaptation method is based on the previous segment variable step-size proportionate normalized least mean squares (SVS-PNLMS) algorithm introduced in [1]. The proposed algorithm has low complexity and guarantees a fast convergence. Simulation results reveal that the proposed algorithm outperforms other counterparts like NLMS (Normalized Least Mean Squares) algorithm.

P.M.B. Mahale

Papers

Pattern Classiffication using SVM with GMM Data Selection Training Methode

A Fixed Dimension Modified Sinusoid Model (FD-MSM) for Single Microphone Sound Separation

A New Adaptive Proportionate Algorithm for Improving the Performance of Precorrection in DVB