This paper aims to enhance accuracy and robustness in audio recognition through the use of the Discrete Wavelet Transform (DWT) with Daubechies wavelets for song identification. This work is important because the actual song identity can be accurately matched based on small music fingerprints that are essential for applications such copyright detection and music identification apps. This method is based on decomposing the frames of a audio sample into sub-bands using Daubechies wavelets and extracting statistical features to form a 8-bit finger-print. This fingerprint is much more compact than that yielded by other techniques such as FFT (256 bits) and FrFT (150 bits). By comparing these fingerprints to a database, the system can accurately identify songs from short snippets even in loud environments without needing song metadata. The results show that the accuracy of Daubechies wavelet method stands at 98% trailing behind existing wavelet based methods where they were reported to have accuracies of 86.7%, 97% and 90%. The high precision and reduced fingerprint size footprint makes the approach ideal for fast, accurate song matching. The novelty of the proposed work lies in its capability to generate very accurate as well as compact fingerprints which overcomes the limitations of previous techniques (e.g. FFT, FrFT etc.). The proposed approach outper-forms previous works in the literature on audio fingerprinting for short audio content and noisy environments.