Indonesian speech recognition grammar using Kinect 2.0 for controlling humanoid robot

摘要

Speech recognition is commonly used to operate a device or perform commands without having to use a keyboard or mouse. Today, this has done on a computer with automatic speech recognition software programs. This paper presents implementation of Indonesian speech recognition grammar using Kinect 2.0 for controlling humanoid robot. Speech recognition grammar done by creating a combination of English lexicon grammar into Bahasa Indonesia language model. The systems successfully recognize speech recognition from different distance (0.5 m, 1.0 m, and 1.5 m) and people (ethnic, age, gender and accent). The average accuracy of speech recognition rate is 93.02%. The greater distance between the Kinect to the speaker, then the accuracy of the speech recognition decreased. We implement Indonesian speech recognition for controlling the humanoid robot and the average accuracy is 94.45%.