Intelligent equalization and compression methods in audio mastering based on machine learning

The article addresses the problem of automating audio mastering using machine learning and digital signal processing methods. The study is relevant due to the rapid growth in the volume of audio content and the need for affordable professional sound processing tools for independent musicians and small studios. A hybrid system is proposed that integrates a music genre classifier (a support vector machine with mel-frequency cepstral coefficients) and a digital signal processing module including equalization, compression, high-pass and low-pass filtering, as well as loudness normalization. The classifier was trained on the GTZAN dataset, while the processing parameters were extracted from the MUSDB18-HQ dataset containing vocal tracks, based on the analysis of root-mean-square amplitude and energy distribution across frequency ranges. Experimental testing on a sample of fifteen tracks of five genres showed that the use of MUSDB18-HQ increases genre classification accuracy from 46.6 % to 60.0 % and improves the signal-to-noise ratio by 2.2 dB compared to the model trained only on GTZAN. Comparison with a commercial counterpart shows that the developed system lags behind in terms of objective metrics; however, the system has advantages: complete transparency of processing parameters due to a detailed log, ease of use, and no usage fee. The novelty of the work lies in the integration of an openly available genre classifier with a digital signal processing module and the adaptation of mastering parameters for vocal tracks. The practical significance lies in the creation of an accessible prototype for educational purposes and for processing demo recordings. Future work includes replacing the support vector machine with neural-network architectures and expanding support for audio formats.

1. Katz R.A. Mastering Audio: The Art and the Science. Burlington: Focal Press; 2014. 408 p.

2. Izhaki R. Mixing Audio: Concepts, Practices, and Tools. Focal Press; 2023. 604 p. https://doi.org/10.4324/9781003303077

3. Sterne J., Razlogova E. Tuning sound for infrastructures: artificial intelligence, automation, and the cultural politics of audio mastering. Cultural Studies. 2021;35(4-5):750–770. https://doi.org/10.1080/09502386.2021.1895247

4. Islam M.A., Uddin M.A., Lee Y.-K. A Distributed Automatic Video Annotation Platform. Applied Sciences. 2020;10(15):5319. https://doi.org/10.3390/app10155319

5. Mycka J., Mańdziuk J. Artificial intelligence in music: recent trends and challenges. Neural Computing and Applications. 2025;37(2):801–839. https://doi.org/10.1007/s00521-024-10555-x

6. Wanasooriya W.M.G.M., Edirisingha R.U.R., Malsha A.G.P.H., Karunatilake C., Amarakeerthi S. Genre-based automatic audio mastering using machine learning and audio signal processing. In: The Second International Conference on Innovation and Emerging Technologies (ICIET 2022), 25–26 November 2022, Virtual Event. 2022. P. 182.

7. Tzanetakis G., Cook P. Musical genre classification of audio signals. IEEE Transactions on Speech and Audio Processing. 2002;10(5):293–302. https://doi.org/10.1109/TSA.2002.800560

8. McFee B., Raffel C., Liang D., et al. librosa: Audio and music signal analysis in Python. In: Proceedings of the 14th Python in Science Conference (SciPy 2015), 06–12 July 2015, Austin, TX, USA. 2015. P. 18–24.

9. Steinmetz Ch.J., Uhle Ch., Everardo F., et al. Audio Signal Processing in the Artificial Intelligence Era: Challenges and Directions. Journal of the Audio Engineering Society. 2025;73(7/8):406–428. https://doi.org/10.17743/jaes.2022.0209

10. Melechovsky J., Mehrish A., Roy A., Herremans D. SonicMaster: towards controllable all-in-one music restoration and mastering. arXiv. URL: https://arxiv.org/abs/2508.03448 [Accessed 10th April 2026].

11. Cobos M., Ahrens J., Kowalczyk K., Politis A. An overview of machine learning and other data-based methods for spatial audio capture, processing, and reproduction. EURASIP Journal on Audio, Speech, and Music Processing. 2022;2022(1):10. https://doi.org/10.1186/s13636-022-00242-x

12. Canyakan S. Machine learning in audio mastering: a comparative study. Journal for the Interdisciplinary Art and Education. 2025;6(1):47–65. https://doi.org/10.5281/zenodo.15074948

13. Cortes C., Vapnik V. Support-vector networks. Machine Learning. 1995;20(3):273–297. https://doi.org/10.1007/BF00994018

14. Hastie T., Tibshirani R., Friedman J. The Elements of Statistical Learning: Data Mining, Inference, and Prediction. New York: Springer; 2009. 745 p.

15. Oppenheim A.V., Schafer R.W., Buck J.R. Discrete-Time Signal Processing. Upper Saddle River: Prentice Hall; 1999. 870 p.

16. Vincent E., Gribonval R., Févotte C. Performance measurement in blind audio source separation. IEEE Transactions on Audio, Speech, and Language Processing. 2006;14(4):1462–1469. https://doi.org/10.1109/TSA.2005.858005

17. Giannoulis D., Massberg M., Reiss J.D. Digital dynamic range compressor design – A tutorial and analysis. Journal of the Audio Engineering Society. 2012;60(6):399–408.

18. Reiss J.D., McPherson A.P. Audio Effects: Theory, Implementation and Application. Boca Raton: CRC Press; 2015. 376 р.

19. Pedregosa F., Varoquaux G., Gramfort A., et al. Scikit-learn: Machine learning in Python. Journal of Machine Learning Research. 2011;12(85):2825–2830.