Human voice extracted biometric features: What can be used for

Strength of encryption algorithm is based on statistical performance of the bits key stream. Generate high quality key stream is a difficult task, which decides the level of security provided by the encryption algorithm. This paper propose two main directions to deal with the biometric features of human voice: first it focuses on how to extract good features that lead to produce encryption keys of lengths 128, 196, and 256 bits that have randomness characteristics, and secondly: how to generate a new encryption keys (18 new key each of 108-bits), only from 12 biometric voice features through a proposed random numbers generator. Statistical performance was performed to make sure that the random key sequence, which was generated in this paper have noise like characteristics to prove the importance of the features that were extracted from the signal of human voice, and how to deal with it. All keys that were used, passed successfully the randomness NIST tests (National Institute of Standards and Technology).