Check regularly for updated readings and assignments.
Dates | Topic | Reading | Assignments |
Aug 28 | Overview of Speech and the Industry |
The Voice in the Machine: Apples and Oranges Open the pod pay doors Hal Demystifying Speech Recognition--the original Demystifying Speech Recognition, Pieracchini's take |
Blog Assignment 1: Speech Application Review |
Sep 1 | Speech Recognizer Components | State of the Art, Makhoul & Schwartz |
Blog Assgn 1: Speech application review due Sept 1 COB |
Sep 4 | Grammars, the Mashup and Speech evaluation | ||
Sep 8 | Interactive applications |
User Interface Design, Bob Morse UI Design, From University of Hawaii |
Sept 9: Audio due |
Sep 11 | Language Modeling | Jurafsky & Martin, Ch. 4 (review) Jurafsky & Martin, Ch. 9.5 |
Baseline (no submission) |
Sep 15 | No class | ||
Sep 18 | More Language Modeling | Goodman, A bit of progress ... | First dev test results due |
Sep 22 | Phonetics and dictionaries | Jurafsky & Martin Ch. 7 |
Final test results due |
Sep 25 | Front End Feature Extraction | Jurafsky & Martin, SLP Ch. 9.1, 9.3 CMU Spectograms, Cepstrum, etc |
|
Sep 29 | Brandeis Monday: No class | ||
Oct 2 | Catching up and project discussions | Which ASR should I choose? Pocketsphinx vs. GoogleSpeech |
|
Oct 6 | HMM Review & VQ, HMMs and Viterbi | Jurafsky & Martin, Ch. 6, Ch.9.2 Andrew Moore, CMU, HMM Tutorial |
|
Oct 9 | Gaussian Mixture Modeling | Jurafsky & Martin 9.3 & 5.5.3 | |
Oct 13 | Eisner Backward-Forward algorithm | HMM Training: Baum-WelchJurafsky & Martin, Ch. 9.4, 9.7 Eisner paper, Eisner spreadsheet |
Perplexity due |
Oct 16 | KALDI | ||
Oct 20 | Kaldi Tutorial | ||
Oct 23 | Another look at Decoding | Jurafsky & Martin, Ch. 9.6 | |
Oct 27 | Advanced Topics in Speech | ||
Oct 30 | Advanced Topics in Speech (cont.) | Cambridge Univeristy TutorialJurafsky & Martin, Ch. 10< | |
Nov 3 | WFSTs in Kaldi, Mirko Hannemann | Kaldi Lectures by Dan Povey | |
Nov 6 | Group work on Advanced Topics and Projects | ||
Nov 10 | Presentations on Advanced Topics | ||
Nov 13 | Continuation of presentations; Kaldi code and data | Code review due by start of class | |
Nov 17 | Speech Synthesis | ||
Nov 20 | Speech Synthesis Continued | Jurafsky & Martin Ch. 8; Choosing Jibo's Voice | Initial Kaldi Training and test due |
Nov 24 | HMM Speech Synthesis, Project midterm presentations & group work | HMM Speech Synthesis Tutorial | |
Nov 27 | No class | ||
Dec 1 | OUCH: Outing the unfortunate characteristics of HMMs | MVP due | |
Dec 4 | Final presentation practice & group work | ||
Dec 8 | Deep Neural Nets | ||
Dec 16 10-12 |
Final Presentations | Presentation Requirements Group reflections |