Abstract

Voice is a fundamental way to communicate with people in a natural atmosphere where we come across many distortions. Speaker identification is a new boon in forensic science which is essential to identify a specific speaker and that a voice cannot be changed and it will prove that it belongs to a single individual. Some voices are naturally or accidentally distorted whereas some are intentionally distorted to disguise the identity of the speaker. The disguised or distorted voices give different values than the authentic ones. The voices can be accidentally disguised by natural environment, by being in a hot or cold atmosphere or deliberately by changing the accent, by keeping hand on mouth, by pulling cheeks, by creating nasal voice etc. The analysis of these voice samples is done by examining using software like Gold Wave, Praat and SSL (speech sound lab). The software help us to examine the voice samples right from extracting clue words to their spectral analysis which are known as spectrograms. Calculating the hash values of the samples provide another authentication to the original samples. Hash value is an alpha-numeric value which gives unique identity to the samples. Hash value has different algorithms but MD5(Message digest) and SHA1(Secure hash algorithm) are more reliable and secured, SHA1 being even more secured than MD5. The differences are made between the samples by looking at the pitch and intensity of the voice of the speakers. The pitch of the two voice samples of the same speaker can also be different because of the natural variation present in the speaker’s voice.

• 1.   Magdin M, Sulka T, Tomanová J, Vozár M. Voice analysis using PRAAT software and classification of user emotional state. Int J Interact Multimed Artif Intell. 2019;5(4):11-9.
• 2.   Ladefoged P, Johnson K. A course in phonetics. 6th ed. Boston: Wadsworth; 2011.
• 3.   Dunn HK. Methods of measuring vowel formant bandwidths. J Acoust Soc Am. 1961;33(12):1737-46.
• 4.   Kersta LG. Amplitude cross‐section representation with the sound spectrograph. J Acoust Soc Am. 1948;20(6):796-801.
• 5.   Bhuta T, Patrick L, Garnett JD. Perceptual evaluation of voice quality and its correlation with acoustic measurements. J Voice. 2004;18(3):299-304.
• 6.   Hirano M, Hibi S, Yoshida T, Hirade Y, Kasuya H, Kikuchi Y. Acoustic analysis of pathological voice: some results of clinical application. Acta Otolaryngol. 1988;105(5-6):432-8.
• 7.   Choudhary A, Chauhan RS, Gupta MG. Automatic speech recognition system for isolated & connected words of Hindi language by using Hidden Markov Model Toolkit (HTK). Int J Adv Res Comput Sci Softw Eng. 2013;3(7):108-13.
• 8.   Selvakumari NAS. Acoustic analysis for human voice disorder classification using optimization and machine learning techniques [thesis]. Coimbatore: Avinashilingam Institute for Home Science and Higher Education for Women; 2019.
• 9.   Marks II RJ. Handbook of Fourier analysis & its applications. Oxford: Oxford University Press; 2009.
• 10.   Körner TW. Introduction. In: Fourier analysis. Cambridge: Cambridge University Press; 1988. p. 221-5.
• 11.   Spagnolini U. Line spectrum analysis. In: Statistical signal processing in engineering. Hoboken (NJ): Wiley; 2017. p. 347-68.
• 12.   Tahilramani N, Bhatt N. Information hiding in line spectrum pair feature of non-voice part of speech signal. In: 2017 International Conference on Smart Technologies for Smart Nation (SmartTechCon); 2017 Aug 17-19; Bengaluru, India. IEEE; 2017. p. 1-6.