A
Atsushi Nakamura
Researcher at Nippon Telegraph and Telephone
Publications - 190
Citations - 3211
Atsushi Nakamura is an academic researcher from Nippon Telegraph and Telephone. The author has contributed to research in topics: Hidden Markov model & Acoustic model. The author has an hindex of 25, co-authored 190 publications receiving 3116 citations. Previous affiliations of Atsushi Nakamura include Nagoya City University & University of Tsukuba.
Papers
More filters
Journal ArticleDOI
Discriminative Training for Large-Vocabulary Speech Recognition Using Minimum Classification Error
TL;DR: This article reports significant gains in recognition performance and model compactness as a result of discriminative training based on MCE training applied to HMMs, in the context of three challenging large-vocabulary speech recognition tasks.
Journal ArticleDOI
Positional cloning of the rice Rf-1 gene, a restorer of BT-type cytoplasmic male sterility that encodes a mitochondria-targeting PPR protein.
Hiromori Akagi,Atsushi Nakamura,Y. Yokozeki-Misono,A. Inagaki,A. Inagaki,Hidekazu Takahashi,K. Mori,Tatsuhito Fujimura +7 more
TL;DR: A rice Rf-1 gene is found that restores BT-type CMS by applying a positional cloning strategy and nine duplications of R f-1A homologs were found around the RF-1 locus in the Nipponbare genome but they do not restore BT- type CMS based on the lack of co-segregation with the restoration phenotype.
Journal ArticleDOI
Ordered domains in tial coexisting with ti3al in the lamellar structure of ti-rich tial compounds
TL;DR: In this article, the morphology and distribution of ordered domains and their relative orientation in the TiAl phase coexisting with the Ti3Al phase in the lamellar structure of Ti-rich TiAl compounds have been studied by transmission electron microscopy.
Journal ArticleDOI
Efficient WFST-Based One-Pass Decoding With On-The-Fly Hypothesis Rescoring in Extremely Large Vocabulary Continuous Speech Recognition
TL;DR: This paper proposes a novel one-pass search algorithm with on-the-fly composition of weighted finite-state transducers (WFSTs) for large-vocabulary continuous-speech recognition and achieves high-accuracy one- pass real-time speech recognition with an extremely large vocabulary of 1.8 million words.
Journal ArticleDOI
Deformation structures in ti-rich tial polysynthetically twinned crystals
TL;DR: In this paper, room temperature deformation structures in Ti-rich TiAl polysynthetically twinned crystals, which contain only a single grain with the two-phase TiAl/Ti3Al lamellar structure, have been studied for both the easy and hard deformation types by transmission electron microscopy.