PhD course in Speech Synthesis Practical exercises Two types of practical exercises are used within the course: * Computer exercises that will be carried out during the lecture days (see Laboratory exercises ). * A set of two home assignments, to be carried out between the scheduled lecture/seminar days. 1. Concatenation cost in speech synthesis The objective of the exercises is firstly to give the students a less abstract understanding of the underlying theories by having to perform calculations using them, and secondly to familiarize them with alternative problem-solving techniques by assessing the solutions of their peers. Instructions The exercises should be solved and submitted before the second day of lectures (i.e. 20/10). The problems will be peer-corrected and discussed during the closing seminar. The peer-correction process is carried out as follows: * The examiner will assign one student as responsible for leading the solution tutorial for each exercise, based on the submitted solutions. * All other students will be given the exercises of another student to correct on spot, guided by the tutorial and discussions with the examiner and the group. * The corrected exercises will then be collected by the examiner for registration of the performance of both the solver and the corrector, before the exercises are returned to the solver. 1.r Requirements: In order to pass, the student should * submit solutions to the concatenation cost problems and * have solved at least 3 of the 5 problems satisfactorily. * have corrected and given adequate feedback to another student's solutions. 2. Evaluation of state-of-the-art commercial speech synthesis The objective of the exercises is to give the students a deeper understanding of the performance that can be expected from state-of-the-art speech synthesis and possible methods for evaluating text-to-speech. Instructions The evaluation report should be submitted before the closing seminar (i.e. 15/12). 1.r Requirements: In order to pass, the student should * submit an evaluation that contains a well-founded analysis of the strenghts and weaknesses of the evaluated system(s). Course responsible: Olov Engwall, engwall@kth.se, 790 75 65