The algorithm proposed in this paper integrates the concepts of variable frame rate and discriminative analysis bascd on Tanimoto ratio to modify the conventional Viterbi algorithm, in such a way that the steady or stationary signal is compressed, while transitional or non-stationary �,ignal is emphasized through the frame-by-frame searching process. The usefulness of each frame is decided entirely within the Viterbi process and needs not to be the samc for different models. To evaluate this algori thm , we tested a speech database of 9 highly confusable E-set English letters. With 5 state and 6 mixture components, the conventional HMM baseline system only delivered the recognition accuracy of 73.9%. In the preliminary experiment using the algorithm proposed in this paper, the recognition accuracy was increased to 82.5%.