Fingerprint is a vector which uniquely describes the information of audio content. The purpose of the audio fingerprinting system is to recognize the original or perceptually equivalent contents. The performance of the audio fingerprinting system is measured how fast and correct a content is recognized. Therefore, the database search section in the system is a crucial factor to determine the performance of the entire system. By the complexity and the vast amount of information of the audio data, the audio data need to be represented by high-dimensional fingerprints. The search of high-dimensional data has been in difficulties for the exponential increase of the query time according to the dimension. This problem is called \`the curse of dimensionality\`. As one of the solutions to remove the limitation of dimensionality, region-conserving hashing scheme is newly proposed by Indyk et al. By appling region-conserving hashing scheme in the previously implemented audio fingerprinting system, the fingerprint space can be divided evenly and the hashed values of fingerprint can be distributed sparsely. Region-conserving hashing function makes close fingerprints into same hashed value. Expermental results show the improvement of the audio fingerprinting system for the performance of database search.