The Arnold Center For Confluent Media Studies

The Edward H. Arnold Center for Confluent Media is also home to cutting-edge multimedia research. There are several papers published by Arnold Center Manager and PhD candidate, Alexander Iliev. Mr Iliev's research has provided solutions for audio watermarking and emotion recognition in speech.

Audio Watermarking
Investigator: Alexander Iliev
Successful watermarking is a compromise between several performance measures, such as the computational complexity, level of introduced distortion and robustness. Various watermarking techniques have been proposed over the past decade. Depending on the application environment, a particular method may provide good performance on a specific measure at the expense of the remaining measures. For example, in audio copyright management, survivability of the watermark whenever the host signal undergoes a variety of processing such as compression, transmission or conversion to analog, is of paramount importance. On the other hand, in content enhancement where a non-hostile environment is assumed, payload is the primary goal.

Emotion Recognition in Speech
Investigator: Alexander Iliev
The information contained in speech is rich and multi-level. While the primary message may be orthographically transcribed, it contains additional information which conveys the identity, gender, age, and physical and emotional condition of the speaker. The automatic extraction of a speaker's emotional state can be very useful in service industries and security application and as a result it has recently received considerable attention. Emotion is carried in the prosodic patterns of speech. Prosodic transcriptions of intonational patterns are much necessary for the fields of speech recognition and synthesis. Emotion recognition is well used for quality enhancement of synthesized speech as well as in surveillance systems.

Publications