Скачать или смотреть INFOMEC: Speaker Identification of the Electronically Disguised Voices using Semi Automatic method

Poster code S-12
POSTER CATEGORY: Science & Technology
PRESENTER: Palak Aneja
PRESENTER’S AFFILIATION: Raksha Shakti University
Abstract
Background: Voice is said to be unique for every individual and can be used for the identification. Significant intra variation in Speech samples can be seen with the changes in the physical and mental conditions. Disguised Voice Samples are challenge to the Forensic Speaker Identification Expert. Changes in the voice samples can be done by physical constriction and also with the technological interventions. There are speech processing softwares which are freely available and can be used to change voice samples. These softwares add some background variations also they change the frequency distribution in the voice. Methods: This paper focuses on the examination of the voice samples which are processed by the freely available android application which changes the female voice samples to that of male. The examination was done on the OT- Expert 6.0. The Examination of the voices was done by the Aural-Acoustics method. Results: For the examination of the voice samples on acoustics parameters such as fundamental frequency(F0) and Formant Frequencies (F1, F2, F3, F4) for 5 vowels compared with their control recordings showed very noteworthy differences. For Aural examination parameters such as relative pauses, background variations and linguistic features such as delivery of speech, specific pronunciations for words remained similar. Conclusion: With some specific formant frequencies for 5 vowels combined with auditory parameters especially the degradation of background noise for suspecting a case of morphing and then comparing it with possible list of suspects for speaker identification. Therefore, on the basis of some specific aural-acoustics features electronically disguised voices can be identified from amongst the suspected voice samples