Voice command system in multi-room smart homes for assist- ing people in loss of autonomy in their daily activities must face several challenges, one of which being the distant condi- tion which impacts the ASR system performance. This paper presents an approach to improve voice command recognition at the decoding level by using multiple sources and model adap- tation. The method has been tested on data recorded with 11 elderly and visually impaired participants in a real smart home. The results show an error rate of 3.2% in off-line condition and of 13.2% in on-line condition.