A Weighted Discrete KNN Method for Mandarin Speech and Emotion Recognition

摘要: Speech signal is a rich source of information and convey more than spoken words, can be divided into two main groups: linguistic nonlinguistic. The aspects speech include the properties word sequence deal with what being said. nonlinguistic have to do talker attributes such as age, gender, dialect, emotion how it Cues also provided in non-speech vocalizations, laught or cry. investigated this article were those audio-visual speech. In conversation, true meaning communication transmitted not only by content but something said, words are emphasized speaker’s attitude toward perception vocal expressions others vital for an accurate understanding emotional messages (Banse & Scherer, 1996). following, we will introduce recognition recognition, which applications our proposed weighted discrete K-nearest-neighbor (WD-KNN) method speech, respectively. consists steps, feature extraction recognition. chapter, methods system. post-processing, different classifiers weighting schemes on KNN-based recognitions discussed overall structure system depicted Fig. 1. briefly previous researches