Style Transfer of Traditional Chinese Bamboo Flute Music Based on AI Model
Ziqi Qin · 2023
The application of AI technology in music can promote the development of multi-source and diversity of the music style, which can effectively promote the exchange of music culture and the enrichment of the music market. In this paper, we proposed DNN-CycleGAN model to convert audio signals of any length from bamboo flute music features to flute music features. We use a short-term Fourier transform algorithm (STFT) to realize the time-frequency conversion of the waveform music, and then utilize a deep convolutional neural network structure (DNN) to extract the features of the waveform music in the Chinese bamboo flute. Then we structure the CycleGAN model to transfer the bamboo flute music into flute music. The experimental results show that the converted music has a precise rhythm, and the timbre characteristics of the flute and bamboo flute are demonstrated to a certain degree.