Speech recognition in living rooms: Integrated speech enhancement and recognition system based on spatial,spectral and temporal modeling of sounds

Research on noise robust speech recognition has mainly focused on dealing with relatively stationary noise that may differ from the noise conditions in most living environments. In this paper, we introduce a recognition system that can recognize speech in the presence of multiple rapidly time-varying noise sources as found in a typical family living room. To deal with such severe noise conditions, our recognition system exploits all available information about speech and noise; that is spatial (directional), spectral and temporal information. This is realized with a model-based speech enhancement pre-processor, which consists of two complementary elements, a multi-channel speech–noise separation method that exploits spatial and spectral information, followed by a single channel enhancement algorithm that uses the long-term temporal characteristics of speech obtained from clean speech examples. Moreover, to compensate for any mismatch that may remain between the enhanced speech and the acoustic model, our system employs an adaptation technique that combines conventional maximum likelihood linear regression with the dynamic adaptive compensation of the variance of the Gaussians of the acoustic model. Our proposed system approaches human performance levels by greatly improving the audible quality of speech and substantially improving the keyword recognition accuracy.     

Ultrashort X-ray pulse generation using subpicosecond electron linac

As a promising tool for ultrafast material analyses, we propose to utilize the X-ray pulse which may be generated in a quite simple manner using subpicosecond electron linacs. The properties of the X-ray were numerically studied with the EGS4 code. Verification of the X-ray generation was also conducted at the Nuclear Engineering Research Laboratory (NERL) linac and clear diffraction patterns of characteristic X-ray were obtained for typical single crystals.     

Comparison of enteral nutrition with combined enteral and parenteral nutrition in post-pancreaticoduodenectomy patients: a pilot study

Background  

Many clinical studies have demonstrated that early postoperative enteral nutrition (EN) improved the postroperative course. Post-pancreaticoduodenectomy (PD), patients tend to suffer from postoperative nausea, abdominal distention, and diarrhoea, causing difficulty in the introduction of EN. In this pilot study, we investigated the appropriate nutritional mode post-pancreatic surgery.     

New monolithic AGC amplifier stabilising output DC level for 560 Mbit/s optical transmission

A new AGC amplifier stabilising the output DC level is proposed and monolithically integrated using 1 ?m Si-bipolar IC technology. First, it is proven that the proposed AGC amplifier is about ten times superior to the conventional one in regard of output DC level stability. Next, it is confirmed that the IC, which exhibits a 720 MHz bandwidth, 39 dB maximum gain and 60 dB gain dynamic range, is feasible for 560 Mbit/s optical transmission.     

A DC-coupled wide dynamic range AGC amplifier IC for 560-Mbit/s optical transmission

A wide-band high-gain AGC amplifier stabilizing the output dc level against a broad gain variation is proposed and monolithically integrated using high-speed 1-μm Si-bipolar IC technology. The fabricated IC exhibits a maximum gain of 39 dB, gain dynamic range of 44 dB, bandwidth of 800 MHz, and output dc-level fluctuation of 8 mV, and realizes wide dynamic range and direct dc-coupling of the multistage AGC amplifier. Also, in order to examine the feasibility of the fabricated IC, a 1.5-μm-wavelength optical transmission experiment was carried out using DFB-LD and InGaAs-APD. Measured minimum received optical power for an error rate of 10^-9is -40 dBm at 560 Mbit/s and -38 dBm at 1.12 Gbit/s. Optical dynamic range of 30 dB is also achieved by using the fabricated IC and APD.     

A reinforcement learning with switching controllers for a continuous action space

Reinforcement learning (RL) attracts much attention as a technique for realizing computational intelligence such as adaptive and autonomous decentralized systems. In general, however, it is not easy to put RL to practical use. This difficulty includes the problem of designing a suitable action space for an agent, i.e., satisfying two requirements in trade-off: (i) to keep the characteristics (or structure) of an original search space as much as possible in order to seek strategies that lie close to the optimal, and (ii) to reduce the search space as much as possible in order to expedite the learning process. In order to design a suitable action space adaptively, in this article, we propose a RL model with switching controllers based on Q-learning and an actor-critic to mimic the process of an infant’s motor development in which gross motor skills develop before fine motor skills. Then a method for switching controllers is constructed by introducing and referring to the “entropy.” Further, through computational experiments by using a path-planning problem with continuous action space, the validity and potential of the proposed method have been confirmed.     

Single-step synthesis and characterization of single-walled carbon nanohorns hybridized with Pd nanoparticles using N2 gas-injected arc-in-water method

Single-walled carbon nanohorns (SWCNHs) hybridized with palladium (Pd) nanoparticles were synthesized by a single-step gas-injected arc-in-water method (GI-AIW) with a Pd wire inserted inside the anode hole. In the arc zone, carbon and Pd were vaporized simultaneously, leading to the formation of hybrid material of SWCNHs and Pd nanoparticles due to effective quenching. Based on TEM and CO chemisorption analyses, Pd nanoparticles were found to be embedded inside SWCNH aggregates. The size of Pd nanoparticles, determined by X-ray diffraction, was in the range of 3–6 nm when Pd wires with diameters of 0.1 and 0.3 mm were used. Using a Pd wire with a diameter larger than 0.5 mm results in larger Pd nanoparticles which tend to be exposed to the outer surface of the hybrid material. According to thermogravimetric analyses, the weight fraction of Pd nanoparticles is increased by increasing the Pd wire diameter although the yield of Pd nanoparticles decreased. SWCNHs hybridized with dispersed Pd nanoparticles, synthesized with 0.1 mm Pd wire, exhibited strong anti-oxidation resistance with a highly graphitic structure.     

Cooking up an interactive olfactory game display.

It's long been possible to give users outside an actual environment that environment's visual and auditory information and thus contribute to establishing presence. However, we've yet to establish much presence when users require olfactory information - such as in environments focused on foods, flowers, perfumes, or, in some cases, more offensive smells. Recently, several VR researchers have become interested in olfaction and olfactory displays that present smells in virtual environments (VEs). In this article, we describe our interactive olfactory display. One of our development goals is to confirm the assumption that users' interactions with the system increases presence. Thus, we used our interactive olfactory display to develop a cooking game in collaboration with electronic engineers and artists.     

没有环境污染韵金属防腐蚀方法

通过投加（1）不只一种马来酸聚合物或它们的水溶性盐：（2）不只一种马来酸、马来酸酐的共聚物和它们的水溶性盐，以及非离子单烯不饱和单体，处理水，从而控制水中朗格里指数≥1．5并且（Si02）×（CaH）≥2000，其中（SiO2）为溶于水的SiO2浓度，而（CaH）为溶于水的钙硬度。以CaCO2（mg／L）计，防止腐蚀。该试剂为环境友好性药剂。