In this paper we report progress made at LIMSI in speakerindependent large vocabulary speech dictation using the ARPA Wall Street Journal-based CSR corpus. The recognizer makes use of continuous density HMM with Gaussian mixture for acoustic modeling and n-gram statistics estimated on the newspaper texts for language modeling. The recognizer uses a time-synchronous graph-search strategy which i...

The spoken content of audio/visual collections such as TV or radio archives is an information resource of enormous potential. The challenge is to develop methods that will make it possible to browse or search these collections. The experimental results presented in this paper demonstrate that syllable-level transcripts provide an important supplement to conventional word-level transcripts for t...

Effects of various additives on the lipase from Burkholderia cepacia (BcL) immobilized on mixed-function-grafted mesoporous silica gel support by hydrophobic adsorption and covalent attachment were investigated. Catalytic properties of the immobilized biocatalysts were characterized in kinetic resolution of racemic 1-phenylethanol (rac-1a) and 1-(thiophen-2-yl)ethan-1-ol (rac-1b). Screening of ...

In this paper we present a new scoring scheme for speech recognition. Instead of using the joint probability of a word sequence and a sequence of acoustic observations, we determine the best path through a word graph using posterior word probabilities. These probabilities are computed beforehand with a modiied forward-backward algorithm. It is important to note that during the search for the be...

We describe a sub-vector clustering technique to reduce the memory size and computational cost of continuous density hidden Markov models (CHMMs). Acoustic models in modern large-vocabulary, continuous speech recognition systems are typically CHMMs. Systems with 100,000 Gaussian distributions of 40-60 dimensions are common, needing several tens of MB of memory. Computing HMM state likelihoods i...

Evidence is presented that the U(1) gauge group of electromagnetism is not fundamental. The viable candidate SU(2) is used to compute the modified propagation properties of the photon at low temperatures and momenta. As a result of strong screening, the presence of a sizable gap in the spectral power of a black body at temperatures T = 5 · · · 20K and for low frequencies is predicted: A table-t...

Side channel attacks can be effectively addressed at the circuit level by using dynamic differential logic styles. A key problem is to guarantee a balanced capacitive load at the differential outputs of the logic gates. The main contribution to this load is the capacitance associated with the routing between cells. This paper describes a novel design methodology to route a design in which multi...

Cryogenic probes have significantly increased the sensitivity of NMR. Here, we present a compact EPR receiver design capable of cryogenic operation. Compared to room temperature operation, it reduces the noise by a factor of ≈2.5. We discuss in detail the design and analyze the resulting noise performance. At low microwave power, the input noise density closely follows the emission of a cooled ...

We present an implementation for Differential Power Analysis (DPA) that is entirely based on Graphics Processing Units (GPUs). In this paper we make use of advanced techniques offered by the CUDA Framework in order to minimize the runtime. In security testing DPA still plays a major role for the smart card industry and these evaluations require, apart from educationally prepared measurement set...

In this paper we report on automatic recognition of Japanese broadcast-news speech. We have been working on largevocabulary continuous speech recognition (LVCSR) for Japanese newspaper speech transcription and achieved reasonably good performance. We have recently applied our LVCSR system to transcribing Japanese broadcast-news speech. We extended the vocabulary to 20k words and trained the lan...