Acoustic Transducer Modeling and Audio Restoration
Using Linear Prediction and Homomorphic Deconvolution


Evan Shenkman


MATLAB Code

Pavarotti Analysis

For this project, the target audio quality is the sound quality of the below recording of Luciano Pavarotti.

Recording of Luciano Pavarotti



Selected Clip of Pavarotti used for Processing



Caruso Analysis

This recording of Enrico Caruso was most likely recording using a metal horn as the transducer. The sound quality is extremely colored by the system function effects of the horn. To try to remedy these effects, homomorphic deconvolution will be used to remove the horn's impulse response from the Caruso recording

Recording of Enrico Caruso


Selected Clip of Caruso used for Processing


Homomorphic Processing

Resulting Transducer Impulse Response & Spectras

Transducer System Function represented as an All-Pole Transfer Function