Protein secondary structure prediction is an important problem in bioinformatics. Inspired by the success of deep neural networks recently, in this paper, we present a novel deep learning architecture which exploits an integrative synergy of prediction by the encoder, convolutional neural network, residual network, and bidirectional recurrent neural network to improve the performance of protein secondary structure prediction. Our architecture utilizes the attention mechanism to extract sequence features, while considering the long-range dependencies present in amino acid sequences using Bidirectional Gated Recurrent Units (BGRU) to capture contextual features. Additionally, we employ a one-dimensional CNN module to extract local relationships present in the sequence. Our model achieves 73.1% Q8 accuracy on the CullPDB test dataset, 67.1% Q8 accuracy on public benchmark dataset CB513, demonstrating its effectiveness.