Wavelet Region implanting watermark upgrades the security framework in Digital Speech Watermarking

The progressions in innovation helped us to use biometric characteristics to verify the people without having them physically. To limit unapproved people and to encourage approved people different human characteristics are used. Face, unique marks, retina, iris, and DNA are some of the biometrics which is used most often to identify people. For additional security in-person authentication and identification systems, a combination of this biometrics can be utilized. Voice can also be used as a biometric similar to other biometrics. Special equipment and computer systems are required to separate this biometrics. This work helps to enhance security by upgrading the modules and embedding the watermark in the speech signal, in-person verification systems. Speakers are validated based on the watermark present in the speech signal. Energy calculations are performed for detail coefficients to select the coefficients where the watermark has to be implanted. The coefficients with less energy are selected for watermark embedding. Inverse discrete transform is applied on approximation and detail coefficients to produce the watermarked speech. The exhibition of the work is assessed by utilizing the subjective and objective measurements such as peak signal to noise ratio, bit error rate, and perceptual evaluation of speech quality. Peak signal to noise ratio is calculated between unique watermark and separated watermark and, unique and watermarked speech. An arrangement of speech articulations of the speaker during the training stage is used to develop the models for the person authentication system. During testing, structures are removed after original and watermarked speech articulations and are used towards calculate the correctness of the system.