Linguistics 478/578
Speech technology

Handouts:

• syllabus
• survey
• HW#1
• HW#2
• acoustics handout
• wave file of Mike saying [a]
• wave file of Mike saying [i]
• snd2spec.m
• spec.m
• mhfildesignexample.fda
• simplespectrogram.m
• sampling demo
• quantization demo
• LPC demo
• HW#3
• LPC sine wave demo
• tV.wav
• aeC.wav
• ioCa.wav
• torment.wav
• LPC pitch demo
• PSOLA pictures
• PSOLA pitch demo
• pitch extraction function (needed for PSOLA demo)
• dowords.m (for HW#4)
• ed.m (for HW#4)
• HW#4
• digits dataset
• training function for digits dataset
• testing function for digits dataset
• length training function for digits dataset
• length testing function for digits dataset
• filterbank demo
• filterbank training function for digits dataset
• filterbank testing function for digits dataset
• "stand" wave
• "stand" textgrid
• dynamic time warping demo
• dynamic time warping function
• DTW testing code for digits dataset
• HW#5
• logarhythmic scaled spectrum demo
• mel filterbank demo
• cepstrum demo
• LPC coefficients training function for digits dataset
• MFCC training function for digits dataset
• Performance on larger word list for question #1 on HW#4
• HW#6
• nouns.txt for HW#6
• verbs.txt for HW#6
• makeran.m for HW#6
• nounsverbs.m for HW#6
• k-means demo
• vector quantization for digits dataset
• Mixed Gaussian demo
• needed for Mixed Gaussian demo
• HW#7
• hmmtst.m for HW#7
• vq7.m for HW#7
• perceptron demo
• neural net demo
• LSTM demo (Note: this does not work as written for windows. I've submitted a request to mathworks to find out what the problem is and will report back when I know more. For now, switch 'adam' to 'sgdm' on windows.)
• precompiled digits data for LSTM demo
• final HW
• data file for final HW
• N-gram approximation demo
• zipf demo

Readings and resources:

• Corpus for approximation demo
• Corpus for zipf demo

Links:

• Wavenet
• Kaldi course
• Praat software
• Matlab software (sitelicensed for UA folks)