HIDING VOICE DATA IN CENTER DENSITY OF SPEECH SPECTRUM FOR SECURE TRANSMISSION

Rupa Patel, Ruma Dhanawate

Abstract: Speech data is becoming an effective and indispensable way for fast information transmission but associated with it are number of unprecedented threats. The idea of this paper is to present a robust speech watermarking method to realize secure voice data transmission. In this method carrier is transformed into frequency domain. Pre-processed normalized covert voice data undergoes exponential transformation which is then substituted in center density of high frequency, high energy subband of carrier. High frequency components are chosen for embedding watermark for two reasons. First, during transmission carrier might get contaminated with noise and filteration generally suppress low frequency components so embedding in high frequency component will keep watermark intact. Second reason is human ears are less sensitive to high frequencies so slight change in amplitude of high frequency components is imperceptible.. Technique uses frequency masking, invisible, and blind approach also. By applying reverse approach sensitive message can be extracted from the watermarked carrier. For embedding and retrieving watermark secret key have been used. Experimental results have shown that proposed method does not change the size of the cover signal even after embedding, does not degrade the quality of carrier and exhibit vigorous voice data hiding performance. Proposed work can be used in those applications where maintaining integrity and secrecy about the information against intentional or unintentional access is given utmost importance

Keywords: Index Terms- Exponential, Frequency Masking, Musical sequence, Speech watermarking, secure voice data, signal to noise ratio

DOI: https://doi.org/10.15623/ijret.2016.0501034