The lectures contained here are the result of almost 15 years of experience teaching digital signal processing and speech recognition graduate courses to students with diverse backgrounds. In the last twenty years, speech recognition has gone from an arcane research area to one of the hottest interface technologies of the new millenium. This course is designed to provide a unique experience for students: a chance to learn the theory and practice behind this technology. The lectures for this course are bundled with a public domain speech recognition system, data, toolkits, and exercises to enhance the learning experience.

To be knowledgable in the art of speech recognition, one must have a wide range of skills including signal processing, pattern recognition, statistics, linguistics, acoustics, human physiology, and, of course, a command of the borad discipline of computer science including programming and software engineering. Since few students have all these experiences before they attend this course, we must provide some background into these topics on an as needed basis. If you are left with the feeling that your particular special interest area has been glossed over, it is by intention :)

Finally, my approach in teaching this course is to aproach the material from an engineering standpoint. Hence, we begin with the signal processing components of the system, move to statistical modeling, and conclude with language processing issues. Throughout the course, we continually attempt to prove that this is really a course in statistical pattern recognition, and that this technology has numerous applications beyond speech recognition.