Introduction to Speech and Machine Learning

About

Introductory machine learning contains self-contained introduction to elementary supervised and unsupervised learning. Evaluation of binary classifiers (e.g. type-I/II errors, ROC analysis). Introduction to modern deep learning approaches (feedforward, convolutive, recurrent). Basic idea of graph neural networks. Introductory speech processing will include speech as a carrier of linguistic information; basics of speech analysis and feature extraction, and statistical modeling. We will illustrate these ideas through application examples ranging from gender recognition, to voice biometrics (speaker recognition), speech recognition, and detection of spoofing attacks (such as audio-based deepfakes). The course contains also topics that introduce the student to research in this field, stemming from research done at the Computational Speech Group at UEF.

The teaching modes will include offline tasks (to be completed prior to contact teaching); lectures and exercises as part of in-campus teaching.

Course leader

Tomi Kinnunen (tomi.kinnunen@uef.fi)

Target group

Master, Doctoral

Course aim

Learning aims: to learn basics about speech and machine learning, and have a gist on research done in the field of speech processing.

Fee info