Subspace Based Speech Enhancement Using Gaussian Mixture Model

Traditional subspace based speech enhancement (SSE)methods use linear minimum mean square error (LMMSE) estimation that is optimal if the Karhunen Loeve transform (KLT) coefficients of speech and noise are Gaussian distributed. In this paper, we investigate the use of Gaussian mixture (GM) density for modeling the non-Gaussian statistics of the clean speech KLT coefficients. Using Gaussian mixture model (GMM), the optimum minimum mean square error (MMSE) estimator is found to be nonlinear and the traditional LMMSE estimator is shown to be a special case. Experimental results show that the proposed method provides better enhancement performance than the traditional subspace based methods.Index Terms: Subspace based speech enhancement, Gaussian mixture density, MMSE estimation.

Two stage iterative wiener filtering for speech enhancement

We formulate a two-stage Iterative Wiener filtering (IWF) approach to speech enhancement, bettering the performance of constrained IWF, reported in literature. The codebook constrained IWF (CCIWF) has been shown to be effective in achieving convergence of IWF in the presence of both stationary and non-stationary noise. To this, we include a second stage of unconstrained IWF and show that the speech enhancement performance can be improved in terms of average segmental SNR (SSNR), Itakura-Saito (IS) distance and Linear Prediction Coefficients (LPC) parameter coincidence. We also explore the tradeoff between the number of CCIWF iterations and the second stage IWF iterations.

Speech Enhancement using Intra-frame Dependency in DCT Domain

In this paper, we present a new speech enhancement approach, that is based on exploiting the intra-frame dependency of discrete cosine transform (DCT) domain coefficients. It can be noted that the existing enhancement techniques treat the transformdomain coefficients independently. Instead of this traditional approach of independently processing the scalars, we split the DCT domain noisy speech vector into sub-vectors and each sub-vector is enhanced independently. Through this sub-vector based approach, the higher dimensional enhancement advantage, viz. non-linear dependency, is exploited. In the developed method, each clean speech sub-vector is modeled using a Gaussian mixture (GM) density. We show that the proposed Gaussian mixture model (GMM) based DCT domain method, using sub-vector processing approach, provides better performance than the conventional approach of enhancing the transform domain scalar components independently. Performance improvement over the recently proposed GMM based time domain approach is also shown.

GMM basedbayesian approach to speech enhancement in signal/transform domain

Considering a general linear model of signal degradation, by modeling the probability density function (PDF) of the clean signal using a Gaussian mixture model (GMM) and additive noise by a Gaussian PDF, we derive the minimum mean square error (MMSE) estimator.The derived MMSE estimator is non-linear and the linear MMSE estimator is shown to be a special case. For speech signal corrupted by independent additive noise, by modeling the joint PDF of time-domain speech samples of a speech frame using a GMM, we propose a speech enhancement method based on the derived MMSE estimator. We also show that the same estimator can be used for transform-domain speech enhancement.

Network Coding for TCP Throughput Enhancement over a Multi-Hop wireless Network

The poor performance of TCP over multi-hop wireless networks is well known. In this paper we explore to what extent network coding can help to improve the throughput performance of TCP controlled bulk transfers over a chain topology multi-hop wireless network. The nodes use a CSMA/ CA mechanism, such as IEEE 802.11’s DCF, to perform distributed packet scheduling. The reverse flowing TCP ACKs are sought to be X-ORed with forward flowing TCP data packets. We find that, without any modification to theMAC protocol, the gain from network coding is negligible. The inherent coordination problem of carrier sensing based random access in multi-hop wireless networks dominates the performance. We provide a theoretical analysis that yields a throughput bound with network coding. We then propose a distributed modification of the IEEE 802.11 DCF, based on tuning the back-off mechanism using a feedback approach. Simulation studies show that the proposed mechanism when combined with network coding, improves the performance of a TCP session by more than 100%.

Design enhancement of the underwater noise control linings through a parsimonious parametric study

Control of sound transmission through the structure and reflection from the structure immersed in fluid media impose highly conflicting requirements on the design of the carpeted noise control linings. These requirements become even more stringent if the structure is expected to be moving with considerable speed particularly under intense hydrostatic pressure. Numerous configurations are possible for designing these linings. Therefore, in this paper, a few lining configurations are identified from the literature for parametric study so that the designer is provided with an environment to analyze and design the lining. A scheme of finite element analysis is used to analyze these linings for their acoustic performance. Commercial finite element software, NISA®, is used as a platform to develop a customized environment wherein design parameters of different configurations can be varied with consistency checks and generate the finite element meshes using the 8-noded hexahedral element. Four types of designs proposed and analysed here address the parameters of interest such as the echo reduction and the transmission loss. Study of the effect of different surface distributions of the cavities is carried out. Effect of static pressure on different designs is reported.

13C-1H HSQC Experiment of Probe Molecules Aligned in Thermotropic Liquid Crystals: Sensitivity and Resolution Enhancement in the Indirect Dimension

The spectra of molecules oriented in liquid crystalline media are dominated by partially averaged dipolar couplings. In the 13C–1H HSQC, due to the inefficient hetero-nuclear dipolar decoupling in the indirect dimension, normally carried out by using a π pulse, there is a considerable loss of resolution. Furthermore, in such strongly orienting media the 1H–1H and 13C–1H dipolar couplings leads to fast dephasing of transverse magnetization causing inefficient polarization transfer and hence the loss of sensitivity in the indirect dimension. In this study we have carried out 13C–1H HSQC experiment with efficient polarization transfer from 1H to 13C for molecules aligned in liquid crystalline media. The homonuclear dipolar decoupling using FFLG during the INEPT transfer delays and also during evolution period combined with the π pulse heteronuclear decoupling in the t1 period has been applied. The studies showed a significant reduction in partially averaged dipolar couplings and thereby enhancement in the resolution and sensitivity in the indirect dimension. This has been demonstrated on pyridazine and pyrimidine oriented in the liquid crystal. The two closely resonating carbons in pyrimidine are better resolved in the present study compared to the earlier work [H.S. Vinay Deepak, Anu Joy, N. Suryaprakash, Determination of natural abundance 15N–1H and 13C–1H dipolar couplings of molecules in a strongly orienting media using two-dimensional inverse experiments, Magn. Reson. Chem. 44 (2006) 553–565].

Performance Enhancement of the Tunnel Field Effect Transistor using SiGe Source

Due to extremely low off state current (IOFF) and excellent sub-threshold characteristics, the tunnel field effect transistor (TFET) has attracted a lot of attention for low standby power applications. In this work, we aim to increase the on state current (ION) of the device. A novel device architecture with a SiGe source is proposed. The proposed structure shows an order of improvement in ION compared to the conventional Si structure. A process flow adaptable to conventional CMOS technology is also addressed.

Enhancement of Strength of Soft Soils with Fly ash and Lime

Disposal of large quantities of fly ash poses a major environmental problem. To enhance its utilization, fly ash is considered for stabilizing of expansive soft soils. Improving the strength of soil, which is of major importance, depends on the pozzolanic nature of fly ash. Fly ashes with high pozzolanic reactivity are widely used but those with less pozzolanic reactivity are greatly inhibited. As the strength development in natural expansive soil considered in this investigation is very less with different percentages of fly ash, an attempt is made to increase the same by addition of lime along with fly ash. Based on several tests conducted, the optimum lime contents for fly ash and soils are 5% and 8% respectively. The strength of compacted soil with different fly ash contents of 10 to 40% with lime contents of 5% and 8% are determined after curing for different periods. The strength improvement for any soil-fly ash mixture, which is substantial with 5% of lime, is further improved with 8% of lime. The strength of soil-fly ash mixtures with any lime content increases with curing period.

An algorithm for dynamic optimal bandwidth allocation in communication networks

We study the problem of optimal bandwidth allocation in communication networks. We consider a queueing model with two queues to which traffic from different competing flows arrive. The queue length at the buffers is observed every T instants of time, on the basis of which a decision on the amount of bandwidth to be allocated to each buffer for the next T instants is made. We consider a class of closed-loop feedback policies for the system and use a twotimescale simultaneous perturbation stochastic approximation(SPSA) algorithm to find an optimal policy within the prescribed class. We study the performance of the proposed algorithm on a numerical setting. Our algorithm is found to exhibit good performance.