Synthesis Samples of Personal Voices
The following table presents comparison examples of synthetic and natural utterances from users
(mostly with ALS and a few with considerable dysarthria) who have used our system over the years to create a personal voice.
With the exception of ModelTalker Kate, which was recorded and hand-corrected at the Speech Research Laboratory, the synthetic voices above were created by users who downloaded InvTool or MTVR off
the internet and recorded their speech in a home or business environment
using off-the-shelf microphones and basic sound cards in their computers. These recordings were then uploaded and
a synthetic voice was automatically created with no editing
done to the speech files.
As you can hear, there is a range of results that may be expected from ModelTalker, depending on the quality/variability of the original voice, recording conditions, and other factors. It is our goal to address as many of these issues as we can, but there are no guarantees. Some of the problems are difficult to solve and are the focus of ongoing research.