| Category | Difficulty |
|---|---|
| HW | 3 |
| Exams | 4 |
This is a course on Advanced Digital Signal Processing and should typically be taken after having done a course on Signals and Systems (18-290) and/or Digital Signal Processing (18-491/18-691) as well as a course on Probability Theory. The course go fairly deep into the topics mentioned below as well as their applications. Homework problems will ask you to implement a lot of these algorithms, usually on speech or music data, and give you a taste of the applications of the topics covered. The signals examined are usually one dimensional signals.
- Sampling
- Change of Sampling Rate and Efficient Implementations
- Polyphase Filters
- Short Time Fourier Analysis/Synthesis
- Spectral Subtraction and Phase Voboding
- Random/Stochastic Processes
- Power Spectral Density Estimation
- Linear Predictive Coding
- Lattice Filters
- Adaptive Filtering
- Array Processing and Adaptive Array Processing
- Miscellaneous Topics: Homomorphic Signal Processing, Wavelet Representations, Robust Automatic Speech Recognition, etc.
There are 11 Homework assignments. Each of these typically consist of 3-4 mathematical/numerical problems and 2-3 MATLAB programming problems. These are initially fairly easy however there is an increase in difficulty around HW3 (Polyphase Filtering) and HW4 (Phase Vocoding).
There are typically two midterm exams and one final exam. The final exam can end up being lengthy. It will be important to have a good grip on the material taught in class.
- Attend lectures to keep at pace with assignments.
- Start assignments early. This should give you sufficient time to deal with programming bugs and get help from office hours.
- Attend recitations. Recitations typically have big hints on how to solve the homework problems and tips to make MATLAB programming easier.
- Attend office hours. These are extremely useful to help with conceptual questions while doing the homework as well as for help with MATLAB programming sections.
- Read the reference/supplementary material provided with each homework assignment. They typically have formulae that helps you solve problems more easily, or complete MATLAB programming problems more easily. They can also cover some topics that are required to solve the homework but may not be covered deeply in class.