Exam will be CLOSED BOOK, CLOSED NOTES, with NO ELECTRONIC COMPUTING DEVICES

For the exam, you should know...

 How to step through one iteration of the following algorithms:
 - Value of hidden RV in HMM using Filtering
 - Most Likely Sequence through HMM using Viterbi (e.g. pitch tracker)
 - Unsupervised learning using Expectation Maximization (e.g. bag of sweets)

 Short answer:
 - In general terms, what are formants, where do they come from, what are they used for
 - In general terms, how to get a spectrum from a waveform / cepstrum from a spectrum
 - Under what conditions is an array of numbers a probability distribution
 - How to obtain a conditional distribution P(A|B) from a full joint P(A,B,C,D)
 - How to decompose a joint distribution using the chain rule
 - What are the independence assumptions in a given Bayes Net
 - How to obtain P(A|evidence) from a given Bayes Net
 - What is Bayes Rule
 - Under what circumstances would one need discounting or backoff
 - What is entropy, perplexity
 - How to estimate cond. prob. tables from fully-annotated data
 - How to estimate cond. prob. tables from partially-annotated data