Computer Science, Language Processing, Speech Recognition
Carnegie Mellon University
I work in four broad areas: Speech recognition, Audio processing, Neural networks and Privacy/Security for voice processing. In audio processing, I work on noise robustness, computational auditory scene analysis, microphone array processing and source separation. In speech recognition, I work on basic research issues that need to be addressed for better automatic speech recognition. In neural networks, I am interested in specialized architectures for signal processing, learning and information routing.
In addition to all of these topics, a major part of my research is focused on privacy preserving algorithms for speech and audio processing. Speech is possibly the most private form of communication; yet, when a speech processing engine such as a voice authentication system or a speech recognition service operates, it requires complete access to the voice recording. This has security, privacy and economic implications, such as snoopers or malicious systems who may use the data to derive undesired information (e.g. speaker demographics), create fake recordings or impersonate the user. Our team continues to develop various cryptographic and other techniques to deal with these problems.