Some comments on "Complex AM signal model for non-stationary signals" by Sircar and Syali

Recently, Sircar and Syali (Signal Processing 53 (1996) 35}45) introduced the complex AM signal model for analyzing non-stationary speech data. It is assumed that the complex AM model has additive errors, which are independent and identically distributed. They provided an e$cient estimation procedure of the unknown parameters using Prony's equation. We re-analyze both the data sets and it is observed that the error assumptions may not be correct in either of them. In this note, we propose a more general complex AM model with additive stationary errors. We use the standard least-squares estimators to estimate the unknown parameters of both the speech data sets. The error analysis indicates that proposed model assumption is correct. Visually, it seems that our estimation procedure provides a better "t than Sircar and Syali (Signal Processing 53 (1996) 35}45) for both the data sets. 2001 Elsevier Science B.V. All rights reserved.