The invention discloses a music style classification method and device, a
computer device and a storage medium. The method comprises: acquiring a
data set; preprocessing audio in the
data set, and inputting the preprocessed audio into a preset deep
convolutional neural network for training to obtain a trained
network model; preprocessing the to-be-classified audio, and inputting the to-be-classified audio into the
network model to obtain a music style recognition result of the to-be-classified audio; wherein the preprocessing comprises the step of separating a
harmonic sound source and a shockwave sound source of the processed audio; and converting the original sound source, the
harmonic sound source and the
shock wave sound source of the processed audio into spectrograms. According to the music style classification method, the
computer device and the storage medium provided by the invention, the audio is converted into the
spectrogram, the deep
convolutional neural network is trainedby using the
spectrogram, and the to-be-classified
audio frequency is classified and identified by using the trained
network model, so that the high-precision classification of the
audio frequency can be successfully realized, the classification speed is high, and the limitation of manual classification is solved.