The Impact of Filtering on the Calculation of Acoustic Characteristics of Useful Speech Signals

Abstract

In criminal proceedings involving the use of audio recordings as evidence, it is particularly important to prepare audio materials correctly for expert analysis. In particular, it is important to comply with technical requirements for recording quality and source authenticity, as well as to document all stages of filtering, if it was used for preliminary listening.

An algorithm for the use of speech signal filtering in forensic analysis has been proposed, which provides for a clear distinction between the stages of primary processing and expert analysis in order to avoid violations of authenticity. The need to develop uniform methodological approaches to the application of digital filters and standards for the storage of processed and unprocessed copies of recordings has been identified.

Within the scope of the tasks set, the sequence of speech signal processing is considered, in particular, primary cleaning and segmentation, which justifies the expediency of applying filtering exclusively at the preparation stage. The influence of individual filter types is analysed: low-frequency filters effectively reduce hum, but can reduce the clarity of consonants; high-frequency filters improve perception, but reduce the strength of the fundamental tone; bandpass filters provide balance, increasing speech intelligibility in noisy environments, but cut off the extreme frequencies of the spectrum.

The results of the study are of practical importance, as they allow us to clearly define the acceptable limits of filtration use, which prevents potential errors in identifying the speaker or verifying the authenticity of the utterance.

Further research should focus on studying the effect of filtering on the identification characteristics of speech in multi-channel recordings, as well as on adapting methods to the analysis of speech of individuals with speech impediments or accented speech. This approach will help improve the accuracy of phonoscopic experts’ conclusions and the effectiveness of criminal investigations using audio and video evidence.