Predicting the acoustical behavior of noise control elements made of fibrous materials, especially the ones comprised by two types of nonwoven materials with different structural parameters, is a relevant topic in large spaces such as gymnasiums, cinemas, shopping malls, airports and stations. In this paper, we propose a more general prediction method for the noise reduction coefficient (NRC) of the dual layered nonwoven absorber by employing general regression neural network (GRNN). In the experiment section, fifty dual layered nonwoven absorbers are specifically designed by five different melt blown polypropylene and hydroentangled E-glass fiber nonwoven materials. Four structural parameters including thickness, area density, porosity, and pore size of each layer are measured, which are used as the inputs of GRNN. The sound absorption coefficients of each absorber are measured with SW477 impedance tube from 80 to 6300 Hz. The sound absorption average at 250, 500, 1000 and 2000 Hz is used to represent the NRC, which is also considered as an output of GRNN at prediction stage. Finally, the prediction framework is carried out after the desired training set selection and spread parameter optimization of GRNN. The prediction results of 10 test samples show that the prediction method proposed in this paper is reliable and efficient.