“You really need the sounds in the context of every other sound and that makes the database big.”

Overall, the piecing-together process took over two months of work, but the recording is fairly convincing.

CereProc now hopes to use the same system to help people who are losing their voice to create realistic digital-versions.