Privacy-Preserving Speaker Identification Using Gaussian Mixture Models
In this chapter we present a framework for privacy-preserving speaker identification using Gaussian mixture models (GMMs). As discussed in the previous chapter, we consider two parties, the client having the test speech sample, and the server having a set of speaker models who is interested in performing the identification. Our privacy constraints are that the server should not be able to observe the speech sample and the client should not be able to observe the speaker models.
- Atallah MJ, Kerschbaum F, Du W (2003) Secure and private sequence comparisons. In: Workshop on privacy in the electronic society, pp 39–44Google Scholar
- Bimbot F, Bonastre J-F, C Fredoille C, Gravier G, Ivan M-C, Sylvain M, Teva M, Javier O-G, Dijana P-D, Douglas R (2004) A tutorial on text-independent speaker verification. EURASIP J Appl Signal Process 4:430–451Google Scholar
- Campbell JP (1995) Testing with the YOHO CD-ROM voice verification corpus. In: IEEE international conference on acoustics, speech and signal processing, pp 341–344Google Scholar
- Michalevsky Y, Talmon R, Cohen I (2011) Speaker identification using diffusion maps. In: European signal processing conferenceGoogle Scholar
- Reynolds DA, Rose RC (1995) Robust text-independent speaker identification using gaussian mixture speaker models. IEEE Trans Speech Audio Process 3(1):72–83Google Scholar
- Paris S, Madhusudana S (2007) A framework for secure speech recognition. IEEE Trans Audio Speech Lang Process 15(4):1404–1413Google Scholar