Objective Evaluation Of Artistic Voice Based On Acoustic Parameters

With the rapid development of science and technology and the continuous improvement of the quality of human life, people are clearer on the definition of health, especially for laryngeal diseases and voice quality. Making an accurate assessment of the voice quality is not only beneficial to the disease’s diagnosis and the treatment of voice, but also plays a vital role in selecting and cultivating those who have a gift for artistic voice. At present, there are many defects in the subjective evaluation of artistic voice, such as time-consuming, laborious and strong subjectivity.To overcome these defects and to explore the painless voice examination and nondestructive examination of voice disorders and other techniques, a deep research on the process of speech signal and objective evaluation of the artistic voice were made in this paper. With the help of the computer and the way of voice analysis, three acoustic parameters were extracted, including fundamental frequency(F0), the first formant(F1) and the third formant(F3). And the method of using Artificial Neural Network to evaluate the artistic voice was proposed.The main works of this paper are as follows:(1)Preprocessing the artistic voice signal, include pre-emphasis,framing and windowing. And then extracting and analyzing the acousticparameters comprise F0, F1 and F3 of the artistic voice signal which have already been pre-processed.(2)The effect of the acoustic parameters in objective evaluation of art voice were analyzed, and combined with neural networks, proposing the method of artistic voice objective evaluation respectively based on Error Back Propagation(BP) neural network and Learning Vector Quantization(LVQ) neural network.(3)Making a subjective evaluation of the artistic voice, and the art voice samples are divided into different categories. Combine the different parameters with two kinds of methods to evaluate the quality of the art voice, after that, comparing the result of objective evaluation and subjective evaluation.Discussing the validity of the method of objective evaluation. And then analysis the best combination of acoustic parameters and objective evaluation to find a better way to evaluate the art voice more objectively and more effectively.Through the simulation analysis, we found that the method based on acoustic parameters contains F0, F1, and F3 can achieve a better function of artistic voice’s objective evaluation. And, the consistency of BP objective evaluation and subjective evaluation up to 85.2%, while using LVQ neural network evaluation method, the consistency between objective and subjective is up to 88.9%.