In this thesis, we introduced the problem of privacy-preserving speech processing through a few applications: speaker verification, speaker identification, and speech recognition. There are, however, many problems in speech processing where the similar techniques can be adapted to. Additionally, there are other algorithmic improvements that will allow us to create more accurate or more efficient privacy-preserving solutions for these problems. We discuss a few directions for future research below.
- Gentry C (2009) Fully homomorphic encryption using ideal lattices. In: ACM symposium on theory of computing, pp 169–178Google Scholar
- Gentry C (2010a) Toward basing fully homomorphic encryption on worst-case hardness. In: CRYPTO, pp 116–137Google Scholar
- Gentry C (2010b) Computing arbitrary functions of encrypted data. Commun ACM 53(3):97–105Google Scholar
- Kearns M, Tan J, Jennifer W (2007) Privacy-preserving belief propagation and sampling. In: Neural information processing systemsGoogle Scholar
- Lauter K, Naehrig M, Vaikuntanathan V (2011) Can homomorphic encryption be practical? In: ACM cloud computing security workshopGoogle Scholar