—This paper proposes word clustering using word embedding. We used a neural net-based continuous skip-gram method for generating word embedding in continuous space. The proposed word clustering method represents each word in the vector space using a neural network. The K-means clustering method partitions word embedding into predetermined K-word clusters.
The work in this paper concerns a small footprint Acoustic Model (AM) and its use in the implementation of a Large Vocabulary Isolated Speech Recognition (LVISR) system for commanding a robot in the Korean language, which requires about 500KB of memory. Tree-based state clustering was applied to reduce the number of total unique states, while preserving its… (More)