Robust Speaker Identification using power spectral density and pitch
Abstract
Human can express their ideas, emotions, thoughts and knowledge with speech as a medium. Speech communication plays important role in human life. The quality and intelligibility is needed for speech signal for easy and accurate exchange of Information. For machine learning applications and in artificial intelligence it is important to recognize speech with accuracy. This can also play a remarable role in robotics and Industrial robots. Most of the time, speech is corrupted by different noises. Insertion of noises reduces the quality of speech, due to which machines unable to identify it. Noises reduce the quality and intelligibility of speech signal. By properly Digital filtering or by using enhancement algorithms. There are There are so many simple and effective method for speech enhancement like Ideal Binary Mask (IBM),Computational Auditory Sense Analysis based methods, Hidden Markov model, Mean square error methods etc. The proposed method uses power spectral density and pitch based robust speech and speaker identification system.