This study clarifies that timbre features are mainly composed of time-frequency, frequency-domain and cepstrum-domain features, and accordingly constructs a timbre feature dataset, which provides an important data basis for subsequent studies. As the dimensionality of timbre features increases, the correlation between timbre features and instrument labels decreases or becomes redundant. To address this problem, we chose to use principal component analysis to reduce the dimensionality of the timbre feature dataset, thus obtaining the Fisher-PCA and IG-PCA features, based on which we used the HMM classifier to categorize the features, and ultimately designed a timbre recognition model based on the HMM classifier, and used the model to carry out the timbre balance control of Western symphony orchestras and Chinese national orchestras. Difference analysis. The acoustic energy of the bass violin, which is a commonly used instrument in Western symphony orchestras, shows a decreasing trend from low to high frequencies, and the energy is lower when it is above the neighborhood of 21.3 KHz. On the other hand, the energy of the Ma Touqin, one of the instruments commonly used in Chinese folk orchestras, covers a wide range and is more evenly distributed, which indicates that the timbre identification model based on the HMM classifier is able to effectively reveal the differences between the Western symphony orchestra and the Chinese folk orchestra in terms of the control of timbral balance.
