Enhancing Fundamental Frequency Estimation of Multi-Harmonic Signals using Spectral Amplitude Knowledge

Description

Fundamental frequency estimation of periodic signals constitutes an important topic in the signal processing domain, with a considerable number of applications in audio, sonar, or industrial purposes like condition monitoring, etc. While prominent methods like the maximum likelihood (ML) fundamental frequency estimator offer asymptotically efficient solutions, they neglect certain prior knowledge that may be available. This work considers prior knowledge of the spectral amplitudes that can be used to enhance the ML estimator. We therefore derive a ML fundamental frequency estimator that beneficially utilizes this knowledge. Monte-Carlo simulations emphasize the improved performance of the proposed estimator compared to the classical ML fundamental frequency estimator, for cases where the spectral amplitudes are perfectly or even only approximately known.

Period	11 Sept 2025
Event title	33rd European Signal Processing Conference (EUSIPCO 2025)
Event type	Conference
Location	Palermo, ItalyShow on map