A speaker-independent continuous speech recognition system using biomimetic pattern recognition

In speaker-independent speech recognition, the disadvantage of the most diffused technology (HMMs, or Hidden Markov models) is not only the need of many more training samples, but also long train time requirement. This paper describes the use of Biomimetic pattern recognition (BPR) in recognizing some mandarin continuous speech in a speaker-independent manner. A speech database was developed for the course of study. The vocabulary of the database consists of 15 Chinese dish's names, the length of each name is 4 Chinese words. Neural networks (NNs) based on Multi-weight neuron (MWN) model are used to train and recognize the speech sounds. The number of MWN was investigated to achieve the optimal performance of the NNs-based BPR. This system, which is based on BPR and can carry out real time recognition reaches a recognition rate of 98.14% for the first option and 99.81% for the first two options to the persons from different provinces of China speaking common Chinese speech. Experiments were also carried on to evaluate Continuous density hidden Markov models (CDHMM), Dynamic time warping (DTW) and BPR for speech recognition. The Experiment results show that BPR outperforms CDHMM and DTW especially in the cases of samples of a finite size.

Polarization-Independent Micro-Ring Resonator on Silicon-on-Insulator

Polarization-independent laterally-coupled micro-ring resonator has been designed and demonstrated. The origin of the polarization-sensitivity of the photonic wire waveguide (PWW) was analyzed. A polarization-insensitive PWW structure was designed and a polarization-insensitive MRR based on this PWW structure was designed by finite difference time-domain method and was fabricated on an 8-inch silicon-on-insulator wafer. The offset between the resonant wavelengths of the quasi-TE mode and the quasi-TM mode is smaller than 0.15 nm. The FSR is about 17 nm, extinction ratio about 10 dB and Q about 620.

Biomimetic pattern recognition for speaker-independent speech recognition

In speaker-independent speech recognition, the disadvantage of the most diffused technology ( Hidden Markov Models) is not only the need of many more training samples, but also long train time requirement. This paper describes the use of Biomimetic Pattern Recognition (BPR) in recognizing some Mandarin Speech in a speaker-independent manner. The vocabulary of the system consists of 15 Chinese dish's names. Neural networks based on Multi-Weight Neuron (MWN) model are used to train and recognize the speech sounds. Experimental results are presented to show that the system, which can carry out real time recognition of the persons from different provinces speaking common Chinese speech, outperforms HMMs especially in the cases of samples of a finite size.

A practical package technique for FBG-based temperature independent sensor

A practical package technique for temperature independent Fiber Bragg grating sensor is proposed. A uniform strength cantilever with two FBG attached on the upper and lower surfaces was utilized as the key element. By detecting two wavelengths differential output, the applied force can be obtained and temperature effects can be eliminated. Experiment results show the sensor has linear response and output signal uctuates less than 12pm as temperature changes from -10 degrees C to 50 degrees C. The maximum thermal error is less than 0.3% of the full measurement range.

A novel low-power input-independent MOS AC/DC charge pump

This paper presents a novel fully integrated MOS AC to DC charge pump with low power dissipation and stable output for RFID applications. To improve the input sensitivity, we replaced Schottky-diodes in conventional charge pumps with MOS diodes with zero threshold, which has less process defects and is thus more compatible with other circuits. The charge pump in a RFID transponder is implemented in a 0.35um CMOS technology with 0.24 sq mm die size. The analytical model of the charge pump and the simulation results are presented.

Selective and lithography-independent fabrication of 20 nm nano-gap electrodes and nano-channels for nanoelectrofluidics applications

A new method has been developed to selectively fabricate nano-gap electrodes and nano-channels by conventional lithography. Based on a sacrificial spacer process, we have successfully obtained sub-100-nm nano-gap electrodes and nano-channels and further reduced the dimensions to 20 nm by shrinking the sacrificial spacer size. Our method shows good selectivity between nano-gap electrodes and nano-channels due to different sacrificial spacer etch conditions. There is no length limit for the nano-gap electrode and the nano-channel. The method reported in this paper also allows for wafer scale fabrication, high throughput, low cost, and good compatibility with modern semiconductor technology.

Fully lithography independent fabrication of nanogap electrodes for lateral phase-change random access memory application

Submitted by 阎军 (yanj@red.semi.ac.cn) on 2010-06-07T01:33:41Z No. of bitstreams: 1 ApplPhysLett_96_213505.pdf: 1153920 bytes, checksum: 69931d8deb797813dd478b5dd0e292c0 (MD5)

Lithography-independent and large scale fabrication of a metal electrode nanogap

A lithography-independent and wafer scale method to fabricate a metal nanogap structure is demon-strated. Polysilicon was first dry etched using photoresist (PR) as the etch mask patterned by photolithography.Then, by depositing conformal SiO_2 on the polysilicon pattern, etching back SiO_2 anisotropically in the perpendic-ular direction and removing the polysilicon with KOH, a sacrificial SiO_2 spacer was obtained. Finally, after metal evaporation and lifting-off of the SiO_2 spacer, an 82 nm metal-gap structure was achieved. The size of the nanogap is not determined by the photolithography, but by the thickness of the SiO_2. The method reported in this paper is compatible with modern semiconductor technology and can be used in mass production.

Wide-Band Polarization-Independent Semiconductor Optical Amplifier Gate with Tensile-Strained Quasi-Bulk InGaAs

A semiconductor optical amplifier gate based on tensile-strained quasi-bulk InGaAs is developed. At injection current of 80mA,a 3dB optical bandwidth of more than 85nm is achieved due to dominant band-filling effect.Moreover, the most important is that very low polarization dependence of gain (＜0. 7dB),fiber-to-fiber lossless operation current (70～90mA) and a high extinction ratio (＞50dB) are simultaneously obtained over this wide 3dB optical bandwidth (1520～1609nm) which nearly covers the spectral region of the whole C band (1525～1565nm)and the whole L band (1570～ 1610nm). The gating time is also improved by decreasing carrier lifetime. The wideband polarization-insensitive SOA-gate is promising for use in future dense wavelength division multiplexing (DWDM) communication systems.

Learning Object-Independent Modes of Variation with Feature Flow Fields

We present a unifying framework in which "object-independent" modes of variation are learned from continuous-time data such as video sequences. These modes of variation can be used as "generators" to produce a manifold of images of a new object from a single example of that object. We develop the framework in the context of a well-known example: analyzing the modes of spatial deformations of a scene under camera movement. Our method learns a close approximation to the standard affine deformations that are expected from the geometry of the situation, and does so in a completely unsupervised (i.e. ignorant of the geometry of the situation) fashion. We stress that it is learning a "parameterization", not just the parameter values, of the data. We then demonstrate how we have used the same framework to derive a novel data-driven model of joint color change in images due to common lighting variations. The model is superior to previous models of color change in describing non-linear color changes due to lighting.

An Algorithm for Group Formation and Maximal Independent Set in an Amorphous Computer

Amorphous computing is the study of programming ultra-scale computing environments of smart sensors and actuators cite{white-paper}. The individual elements are identical, asynchronous, randomly placed, embedded and communicate locally via wireless broadcast. Aggregating the processors into groups is a useful paradigm for programming an amorphous computer because groups can be used for specialization, increased robustness, and efficient resource allocation. This paper presents a new algorithm, called the clubs algorithm, for efficiently aggregating processors into groups in an amorphous computer, in time proportional to the local density of processors. The clubs algorithm is well-suited to the unique characteristics of an amorphous computer. In addition, the algorithm derives two properties from the physical embedding of the amorphous computer: an upper bound on the number of groups formed and a constant upper bound on the density of groups. The clubs algorithm can also be extended to find the maximal independent set (MIS) and $Delta + 1$ vertex coloring in an amorphous computer in $O(log N)$ rounds, where $N$ is the total number of elements and $Delta$ is the maximum degree.