Now we are in the wide world of the web user driven reviews plays an important role for new consumers/individuals in shopping the products or taking up a new service and analyse the quality of the product/service. There are many e-commerce sites like Trip Advisor, Make My Trip, Trip Factory, Yelp, Zomato basically generate business through user driven online reviews. According to local consumer review survey conducted by Bright Local in the year 2015, 92% consumers read online reviews. 60% the consumers take up call in buying after reading 4-7 reviews. An efficient framework is needed to analyse the user driven content and generate the polarity of the review such as positive, negative and neutral. Hence this paper proposes Review Analysis Framework (RAF), which easily describes the polarity of the product/service for new individuals rather than going through numerous reviews. RAF based opinion mining is based on lexicon based approach. RAF consists of 3 stages: 1). Speech Recognition 2). Preprocessing and feature extraction 3). Polarity shift problems. Speech recognition is performed using Java API. The feature words are extracted using niosto tool and weightage is assigned using sentiwordnet 3.0. Experimental results are carried out using amazon product data datasets which are given to the system using human agent i.e., through speech. RAF performance analysis is performed on various domains like Books, Electronics, Beauty etc., and experimental results are identified and discussed.