Binary codes play an important role in many computer vision applications. They require less storage space while allowing efficient computations. However, a linear search to find the best matches among binary data creates a bottleneck for large-scale datasets. Among the approximation methods used to solve this problem, the hierarchical clustering tree (HCT) method is a state-of the-art method. However, the HCT performs a hard assignment of each data point to only one cluster, which leads to a quantisation error and degrades the search performance. As a solution to this problem, an algorithm to create hierarchical soft clustering tree (HSCT) by assigning a data point to multiple nearby clusters in the Hamming space is proposed. Through experiments, the HSCT is shown to outperform other existing methods.