|SPCH04: Robust pitch tracker for speech processing|
Tracking the fundamental frequency (pitch) of speech is a problem with applications in various areas of speech processing including speech compression, pitch modification, speech recognition for tonal languages such as Mandarin Chinese, using prosodic cues in speech recognition, language learning, ...
The student will examine existing pitch extraction methods for criteria such as robustness to noise, robustness to bandwidth constraints, computational complexity and error rate. Finally, he/she will implement the algorithm and compare it to an existing baseline.
Experience with programming and some background in signal processing techniques are helpful skills.