Instantaneous frequency decomposition: an application to spectrally sparse sounds with fast frequency modulations.

Area of Science:

• Signal Processing
• Auditory Neuroscience

Background:

• Classical time-frequency analysis relies on filter amplitude responses, omitting phase information.
• Instantaneous frequency analysis utilizes phase derivatives for precise frequency calculation, mirroring cochlear encoding.

Purpose of the Study:

• To develop a high-resolution analysis methodology for sparse sounds using instantaneous frequencies.
• To introduce self-validating mechanisms within the analysis to assess signal density and analysis quality.

Main Methods:

• Comparing tonotopic and instantaneous frequency information to select optimal filter positions.
• Employing cross-checks between neighboring channel frequency estimates to optimize filter bandwidth and assess analysis quality.
• Developing a method for optimal time-frequency representation without prior signal information.

Main Results:

• The method decomposes sparse signals into precisely tracked time-frequency contours.
• For spectrally dense signals, inconsistent estimates from neighboring channels signal analysis validity.
• Achieves optimal time-frequency representation adaptable to signal sparsity.

Conclusions:

• The described methodology offers high-resolution analysis of sparse sounds via instantaneous frequencies.
• The self-assessment feature allows the method to determine its applicability based on signal characteristics.
• Potential parallels exist between these optimization principles and cochlear encoding mechanisms.