Refined Sound Decomposition Method Mirrors Human Auditory Perception, Opens Up New Possibilities in Sound Processing
Researchers at the Aalto University Acoustics Lab have refined a method to decompose sound into its three basic components: sines, noise, and transients. This model mirrors how humans perceive sound, separating whistles, clicks, and hisses.
Disclaimer: This post was written with ChatGPT.
Source: Acoustics researchers decompose sound accurately into its three basic components.
The team leveraged insights from auditory perception, fuzzy logic, and perfect reconstruction in their research. In the refined method, sines and transients cannot belong to the same class at the same time, but each can coexist with noise, providing a more nuanced description of sound.
A listening test showed that this method outperformed previous ones in decomposing most sounds, with the exception of sounds with strong vibrato. Additionally, this decomposition method showed potential in modifying the time-scale of sound, particularly in slowing down music.
The researchers believe this new method will offer exciting possibilities in sound processing. It could lead to improvements in slow-motion audio quality, novel music remixing techniques, and distortion-free dynamic range compression.
