Design of Piano Singing Signal Classification Framework Based on Multi-discrimination Model

Based on the analysis of Fisher's multivariate discriminant model and BP neural network model, this paper designs a classification framework for piano playing and singing signals. First of all, this article introduces the piano's notes, tones and timbre, and understands the piano's twelve temperaments and notes. Next, preprocess the data of the piano playing and singing signal, including sampling and quantization, and then perform feature extraction on the signal. Finally, adopt the Fisher multi-discrimination model and BP neural network model to classify the characteristic signals. Experimental results show the effectiveness of the classification framework.