ECSE-570: Automatic Speech Recognition

Syllabus

Calendar

Lectures

Assignments

This course in Automatic Speech Recognition (ASR)  is being offered under ECSE-570.     The course includes lecture slides along with  tutorial examples for  modeling, classification, and recognition as applied to the speech problems presented in the course.

Continuing advances in the field of automatic speech recognition (ASR) have lead to new mechanisms for human-machine interaction and new means for the interpretation and understanding of multimedia information.   ASR is the core technology that underlies systems for human-machine dialog, automatic dictation, transcription of multimedia broadcasts, and human-computer interfaces on a growing array of mobile devices.  This course will address theoretical foundations, essential algorithms, and experimental strategies related to ASR.   It will draw heavily from methods in machine learning and signal processing, considering how these methods relate to human speech production and speech perception - and how they perform in various applications.

The content of the course is divided into three parts.  The first part provides a basic background in acoustic phonetics and signal representations.  The second and major part of the course develops pattern classification, stochastic modeling, language modeling, and search algorithms as they
are applied in ASR.  Finally, the third part develops advanced techniques that include making
ASR more robust against sources of variability, integrating ASR with other user interface modalities, and the role of ASR in speech understanding.