GSLT: Speech Technology 1

Fall semester 2004

Nordic Graduate School of Language Technology
Graduate School of Language Technology

Back to Speech Technology 1 home page

Practical assignment – Speech Synthesis


Experiments on concatenative synthesis in a very limited domain.


We suggest that you all do a "system" for a limited domain, e.g. a number reader from 1 to 1000. This minimally includes: Selection and recording of the units, example of concatenations with some small evaluation. This can be done "manually" in wave surfer.


For a higher ambition, if you are computationally up to it, a more automatic system may be built.


If you work in groups which is encouraged, more interesting aspects can be investigated, including unit size, possibility for contrastive stress of expressive synthesis.


Result and experiences from the work shall be presented at the Closing Seminar.


Björn Granström