Sound-Recognition Method for Helping Us Respond Appropriately to Sounds in Daily Life
Mamoru Iwaki, Shun Nakayama · 2018
Sounds convey useful information for daily life and should be recognized incrementally as each portion of a sound arrives for fluent communication. Speech transmits not only verbal information but also non-verbal information, e.g., speaker individuality and emotion. Environmental sounds help us sense the circumstances around us. Usually, such sounds help us prepare for our next behavior. In this paper, we propose a progressive method for sound recognition that is suitable for the above situation. We applied the method to recorded sound data to evaluate its performance. The results indicate that its recognition speed was fast, but its recognition accuracy should be improved for practical use.