In today's world, human communication operates increasingly with technical equipment. For example in Finland the use of mobile phones is quite popular in every age group and among both sexes. In addition to private communication people increasingly use technical devices in handling their daily affairs also e.g. using the telephone banking instead of literally going to bank. As consequence the importance of saving the spoken information from any significant communication in forensic or legal/juridical purposes has naturally grown. Recorded conversation can be used as evidence in court room.

For economic reason compressed speech (or any kind of) data is very popular nowadays. Due to this most of the forensic case work deals with the study of compressed, filtered and in different ways altered speech material. And from forensic point of view, this otherwise brilliant technical discovery does not - unfortunately - help the forensic speech and audio analysis. In worst cases a lot of valuable information is lost when the speech signal is heavily compressed.

Compression of the speech signal is a difficult area of forensic studies. It can be said that the more the speech undergoes different modifications the less it resembles its original representation. This has naturally consequences in the reliability of the quantitative analysis and to the evaluation whether the differences between the samples are intra- or inter-individual. The aim of this study is to find out the effects of certain compression algorithms in forensic speech analysis.

Esitelmä liittyy Puhujantunnistushankkeeseen osana Suomenkielisen puheteknologian yhteishanketta (SuoPuhe TEKES 40285/00, 40406/01, 40238/02,HY 460325).