EECS 225d (3 units)
Audio Signal Processing in Humans and Machines
The focus of this recently modified course is on engineering models for speech and audio processing. Mirroring the new edition of the associated textbook, the class will include more material about current audio processing methods such as psychoacoustic audio coding (e.g., MP3) and sound source separation. The methods discussed are used to design systems for analysis, synthesis, and recognition of speech and music. For many of these topics we will discuss not only the engineering methods, but also some of the physiological and psychoacoustic properties of the human auditory and speech generation systems. This latter information can provide an important perspective: how can we make use of knowledge about these natural systems when we design artificial ones?
Topics will include, among others: auditory physiology; room acoustics; music signal analysis; speech synthesis; models of speech production and perception; signal processing for speech analysis; robustness to environmental variation in speech recognizers; statistical speech recognition, including introduction to Hidden Markov Model and discriminative approaches.
Previous classes have included students from both EE and CS, whose varied backgrounds have enriched the course.
Prerequisites: EE123 or equivalent, and Stat 200A or equivalent; or grad standing and consent of instructor
Text: Gold, Morgan, and EllisSpeech and Audio Signal Processing, Wiley Press, 2nd edition, 2011.