In the presence of ambient noise, speech communication is prone to strong background interference, resulting in a degradation of voice signal quality and intelligibility. To address these challenges and enhance the performance of speech communication in ambient noise, this paper proposes a novel dual-path Transformer architecture incorporating an efficient channel attention mechanism. The proposed attention mechanism computes the importance weight for each frequency bin and dynamically selects relevant frequency bins to improve information transfer and interaction across different frequency components. The forward network of the dual-path Transformer utilizes BiGRU and GeLU functions as sequence processing and activation functions, respectively, leading to improved extraction of speech context information. The performed experiments utilizing the Voice Bank + DEMAND dataset showcase the superior performance of speech enhancement compared to existing systems.