Here you will find some practical exercises to help you understand the material in the courses.
In these simple exercises, we get our hands on speech and other audio signals, and analyse them in various ways. We use the Wavesurfer and Praat tools.
Festival is a widely used research toolkit, so we investigate how it works, step-by-step. It is not perfect, and we discover various types of errors it makes, then understand why they occur.
A simple but functional digit recogniser built from scratch: record and label data, train HMMs, create a language model, and recognise the test data. Extend to other speakers & digit sequences.
Although it's not really used anymore, Dynamic Time Warping (DTW) is a nice introduction to the key concept of Dynamic Programming.
Record your speech and build a unit selection voice for Festival. Create variations of the voice, add domain specific data, or vary the database size. Evaluate with a listening test.
This exercise assumes that you have already built your own unit selection voice, and therefore have all the data you need.
Special instructions for the unit selection and DNN labs for this summer school.