论文部分内容阅读
本文综述了当前语音识别理解的发展趋势和最新进展。指出美国在不依说话人的大词汇表的连续语音隐马尔柯夫模型识别方面起主导地位，日本在大词汇表的连续语音神经网络识别、模拟人工智能进行语音后处理方面起主导地位，并介绍了目前国际上最优秀的语音识别理解系统。
This article summarizes the current development of speech recognition and the latest development. It points out that the United States plays a dominant role in continuous speech hidden Markov model recognition of non-speaker-based vocabularies. Japan plays a leading role in continuous speech neural network recognition of large vocabularies and analogue artificial intelligence in speech post-processing, and introduces At present, the world’s best voice recognition understanding system.