Abstract: Formant is one of important features for speech signal. It is very significant to enhance the speech intelligibility (SI) for deafness patients. At present, the commonly used loudness compensation algorithms for digital hearing aid, for instance multi-channel loudness compensation (MCLC) and wide dynamic range compression (WDRC), have strongly negative effects to the formant and the results are poor for patients’ SI. In this study, a new MCLC algorithm based on formant extraction combining formant detection method is presented here. The algorithm can play protective effect for formant by redesigning of filter bank and joining a formant extraction module. Results show the proposed algorithm can have a suitable compensation effect for 4 common deafness types while achieving more excellent performance than MCLC and WDRC on protecting formant structural integrity.