Towards Automatic Sign Translation PDF Download

This book provides the readers with retrospective and prospective views with detailed explanations of component technologies, speech recognition, language translation and speech synthesis. Speech-to-speech translation system (S2S) enables to break language barriers, i.e., communicate each other between any pair of person on the glove, which is one of extreme dreams of humankind. People, society, and economy connected by S2S will demonstrate explosive growth without exception. In 1986, Japan initiated basic research of S2S, then the idea spread world-wide and were explored deeply by researchers during three decades. Now, we see S2S application on smartphone/tablet around the world. Computational resources such as processors, memories, wireless communication accelerate this computation-intensive systems and accumulation of digital data of speech and language encourage recent approaches based on machine learning. Through field experiments after long research in laboratories, S2S systems are being well-developed and now ready to utilized in daily life. Unique chapter of this book is end-2-end evaluation by comparing system’s performance and human competence. The effectiveness of the system would be understood by the score of this evaluation. The book will end with one of the next focus of S2S will be technology of simultaneous interpretation for lecture, broadcast news and so on.