Compact Speech Decoder Based on Pure Functional Programming
Current speech recognition systems are implemented using very large and complex programs. This makes difficult to learn such systems and test a new idea in speech recognition that requires program modification. Therefore, compact system suitable for educational and prototyping purposes is required. We developed a very compact speech decoder based on pure functional programming, which has been studied in software engineering as a means to describe complex systems in a highly ed manner. The decoder is based on weighted finite state transducer and is described using less than 400 lines of codes. The performance of the decoder is demonstrated with large vocabulary continuous speech recognition experiments using a trigram language model and a cross-word triphone hidden Markov model.
Takahiro Shinozaki Masakazu Sekijima Shigeki Hagihara Sadaoki Furui
Chiba University, Chiba, Japan Tokyo Institute of Technology, Tokyo, Japan
国际会议
2011亚太信号与信息处理协会年度峰会(APSIPAASC 2011)
西安
英文
1-4
2011-10-18（万方平台首次上网日期，不代表论文的发表时间）