This paper presents a fuzzy system to discriminate speech signals from background. In our previous works, we had developed a method for speech classification. For speech classification, the universe of discourse is divided into two types, and each type is treated as a class. These are background and speech signals. The rectangular fuzzy system (RES) is used to classify frames and integrate the rule-based approach. The variance of first detail and the third approximation can extract fuzzy classification rules. Experimental results demonstrate the superior performance to the conventional ones. The effectiveness of the proposed system is confirmed by the experimental results.