PhonotacticReconstructionofEncryptedVoIPConversations.PDFVIP

Phonotactic Reconstruction of Encrypted VoIP Conversations: Hookt on fon-iks Andrew M. White Austin R. Matthews Kevin Z. Snow Fabian Monrose Department of Computer Science Department of Linguistics University of North Carolina at Chapel Hill Chapel Hill, North Carolina amw, kzsnow, fabian@, armatthe@ Abstract—In this work, we unveil new privacy threats against ciphers for encryption—interact to leak substantial infor- Voice-over-IP (VoIP) communications. Although prior work mation about a given conversation. Specifically, researchers has shown that the interaction of variable bit-rate codecs and have shown that this interaction allows one to determine the length-preserving stream ciphers leaks information, we show that the threat is more serious than previously thought. In par- language spoken in the conversation [55], the identity of ticular, we derive approximate transcripts of encrypted VoIP the speakers [2, 41], or even the presence of known phrases conversations by segmenting an observed packet stream into within the call [56]. subsequences representing individual phonemes and classifying Rightfully so, critics have argued that the aforementioned those subsequences by the phonemes they encode. Drawing on threats do not represent a significant breach of privacy. For insights from the computational linguistics and speech recog- nition communities, we apply novel techniques for unmasking example, the language