Research on Chinese Art Song Recognition Based on Optimized Deep Belief Network Algorithm

With the increasing popularity of traditional Chinese music, recognizing Chinese art songs, particularly those based on classical poetry, has become an important research topic in the field of music information retrieval. Traditional methods for recognizing Chinese art songs often struggle with issues such as high complexity in musical structure and insufficient representation of tonal nuances. This paper proposes an optimized Deep Belief Network (DBN) algorithm for the recognition of Chinese art songs based on classical Chinese poetry. The method first extracts relevant audio features, such as pitch, rhythm, and timbre, and then applies an optimized DBN model to improve the recognition accuracy. The optimization technique incorporates a combination of feature selection and fine-tuning of network parameters to enhance the model’s ability to distinguish subtle musical characteristics. Experimental results demonstrate that the proposed approach significantly outperforms conventional methods in terms of recognition accuracy and robustness, particularly when handling the rich linguistic and melodic structures inherent in Chinese art songs. This research provides a novel solution for automatic recognition and classification of Chinese art songs and offers potential applications in cultural heritage preservation and intelligent music systems.