After 300 CPU-hours of intense computation, a new recording script for Arpasing has been generated.
Unlike the lists you've seen before, this one consists of actually meaningful words and phrases selected from a small subset of public-domain books from Project Gutenberg, yet it is different from CMU Arctic (which is also based on Gutenberg) in that the new Arpasing script is designed for singing synthesis. You can take a glimpse at the first few lines,
was Matthew's consolatory rejoinder
he stood before the wide opening
we shall begin our researches here
with his usual craving on him
but this little line of dancing men
Theoretically, this new script can match the current 220-line script performance-wise, using its first 65 lines only (though the total number of syllables remains roughly the same); on the top of that, the rest 135 lines should double the coherency.
But speech synthesis is all about experiments. Before rushing into the recording studio, I'd like to focus on improving the ease of pronunciation by taking out phrases that are hard to read even to native speakers. To keep the selection as unbiased as possible, I kindly request users of Arpasing and people with an interest in the matter to review and comment on the draft:
Thanks for your support!
(view the original post on my blog)