Speech accent identification with vocal tract variation trajectory tracking using neural networks

Identifying the accent of a speaker can improve the performance of speech recognition systems. Furthermore it can be a useful tool in many other areas such as forensic speech analysis. Speech patterns are proving to be increasingly valuable in criminal investigation. This paper proposes a method for speech accent identification that uses a scaled conjugate gradient back-propagation learning neural networks to generalize the accent dependent temporal variations of speaker´s vocal tract. Automatic accent identification has become a serious consideration and also a challenge in modern automatic speech recognition, processing and analysis systems