This study, considers an analysis of the 1996 Centennial Olympic Park Bombing 911 phone call with respect to several speech processing issues. First, since background noise is present in the recording, five previously formulated speech enhancement algorithms are compared for quality improvement. Some of these include: nonlinear spectral subtraction, MMSE filtering, and auto-LSP constrained iterative Wiener filtering. Afterwards, the problem of automatic speaker height estimation from voice data is considered. Finally, the implications that such an analysis could have on extracting further information from the 911 call are discussed.
展开▼
