Music is created, streamed, and consumed through electronic platforms and while this format does not suffer the pops and crackles or the wow and flutter of vinyl or magnetic tape, there is always the persistent problem of noise interference and how to minimise it. Taoru Kong of Xi’an Siyuan University in Xi’an, and Yanli Shen of Guangdong Polytechnic of Science and Technology in Zhuhai, China, writing in the International Journal of Computational Systems Engineering, point out that electronic music can be particularly susceptible to the problem of noise. The team is developing a powerful new approach to reducing noise interference.
The team’s system combines hardware and software designed to tackle noise interference. It begins with a music signal acquisition module that captures the raw audio. This data is then processed using an audio codec to refine the signal quality that exploits a large training data set and a clustering algorithm to detect and isolate unwanted noise. By analysing patterns within the music signal, the system can target interference, applying a statistical tool known as a an improved wavelet transform to eradicate the noise without distorting the original music.
In addition to this analytical system, the team has incorporated a so-called autocorrelation filtering algorithm. This step further cleans up the signal by isolating the music. This, the researchers say, produces a remarkably clear audio output, which is critical in a world where the smallest distortion can detract from the listener’s experience.
The experimental results show that this system outperforms traditional noise reduction methods by a significant margin. For example, it achieves a signal-to-noise ratio of 30.12 dB, which is considerably higher than the 27 to 28 dB range seen with conventional techniques.
Kong, T. and Shen, Y. (2024) ‘Design of music signal enhancement system based on big data clustering technology’, Int. J. Computational Systems Engineering, Vol. 8, Nos. 3/4, pp.182–191.