According to the signal theory vocal speech presents a non-stationary random process due to time variation of its main characteristics. A special emotional impact is gained during ensemble and choir performances. The article provides the correlation analysis results and the autocorrelation function of ensemble singing calculated, taking into account the non-stationary nature of acoustic signals. The studies have shown that there is a certain visual resemblance between correlation portraits of a non-professional student ensemble and a non-professional solo singer. At the same time the correlation portrait of student and professional co-performance tends to a professional singer correlation portrait. This helps to make a quantitative assessment of an ensemble’s professionalism which, alongside with a vocal teacher’s subjective assessment, can provide an objective view on singers’ voices condition to optimize the academic voice training process in an ensemble, which is important when using this voice assessing method in music schools, colleges and universities.
Keywords: Vocal speech, acoustic signal, ensemble singing, choir performance, non-stationary random process, autocorrelation function, quantitative assessment
Vocal speech is a random process. Taking into account the fact that the main characteristics of a singing voice vary with time, in the article vocal speech is studied as an acoustic nonstationary random process. To describe this process we should calculate one of its informative parameters – the autocorrelation function. The findings show that there are some differences between the autocorrelation functions for professional and nonprofessional singers. This helps to make a quantitative assessment of a singer’s professionalism which, alongside with a vocal teacher’s subjective assessment, can present an objective view on a singer’s voice and provide a good basis for optimization of academic voice training. Moreover a teacher does not have to be technically educated because this quantitative assessment is presented in diagrams which are quite easy to assess visually. It is especially important to consider when we use the given voice assessing method in music schools, colleges and universities.
Keywords: Vocal speech, acoustic signal, nonstationary random process, autocorrelation function, the maximum autocorrelation interval, quantitative assessment
A voice activity detector (VAD) is a device, which analyses a speech signal and generates the signal corresponding to the period containing only noise. In the present work is offered VAD, which increases the probability of correct detection the presence or absence of human speech.
The correct detection begins with SNR 7-10 dB. The quality of derived speech signal remains the same.
Keywords: voice activity detection, voice activity detector, VAD, silent interval detection, speech, speech signals
Current article presents the results of research, aimed at raising the effectiveness of automatic prosodic analysis of speech. During the research the PC program that performs recognition of seven emotional states using 16 features of speech signal was developed. Effectiveness of recognition of seven emotional states was estimated during the experiment with material of Berlin database of emotional speech. The result showed rising of mean probability of correct recognition, in comparison with known analogs.
Keywords: speech, intonation, emotions, speech
Information about authors of issue №3 (2012)
Information about authors of issue №4 (2011)