| Autor: |
Mane, Vikram, Patil, Dipak, Patil, Shubhangi |
| Předmět: |
|
| Zdroj: |
AIP Conference Proceedings; 2023, Vol. 2717 Issue 1, p1-10, 10p |
| Abstrakt: |
Speech is one of the communication media used by human beings. Human can express their emotions and thoughts in terms of speech. In communication the quality and intelligibility of speech is required. For artificial processing of the signal preserving the information becomes extreme impotents. Analyzing the signals in more understandable form will become remarkable form for Artificial and Machine learning applications. Preserving and identifying quality and intelligence of voice signals will play vital role in voice based appliances like robotics and Industrial robots. Mostly, speech is corrupted by various noises like random noise. It majorly affects to degrade the quality and the loss of information which becomes difficult for machine learning applications to extract meaningful information. Noises reduce the superiority and clearness of speech signal. Various speech enhancement processes have been proposed like Ideal binary Mask, Hidden Markov speech recognition algorithms, or Mean square error methods which may be suitable for CASA (Computational Auditory Sense Analysis) based models. Which are effective in noisy environment also. The proposed method uses a novel method for robust speaker identification which works on power spectral density. To improve the performance of the system a pitch of speech is added parameter considered. The proposed system can become one of the step to help CASA based systems which works like human auditory System. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
