As my tongue and throat muscles have grown weaker, my speech has become more slurred, indistinct and hard to understand. I have already noticed that as a result of speech hesitancy, I have become more withdrawn in social settings, less spontaneous and not as likely to offer a quip, bad pun or telling response.
At some point, unaided verbal communication will no longer be possible.
But even when I lose the ability to speak, I will not lose my voice.
Thanks to remarkable software developed by a dedicated team in Delaware, I will retain the capacity to verbalize with a synthetic voice that is built from my own recordings, and can speak any words, phrases or sentences I type into it.
The software is called ModelTalker and Dr. Tim Bunnell and his colleagues at the Nemours Alfred I. duPont Hospital for Children, along with AgoraNet Inc., a small software company, are developing it. Funding has come from a variety of government agencies and private foundations.
A couple of months ago, I spent a week or so recording about 1,700 short sentences that contain multiple examples of the basic sounds and combinations of sounds that comprise spoken English. From that data, the ModelTalker team and software was able to build me a replacement voice that, while a little robotic, does sound like me, and with a little more fine-tuning, may be close enough to fool my closest friends.
My earlier posts often have led me to the conclusion that paying careful attention to the details and the “small things” has become a critical part of my strategy for staying ahead of my disease’s intents. So it should not be surprising that I have learned how, by focusing on the very fine details of speech sounds and the spaces and relationships between them, the ModelTalker project is able to reconstruct fluid, understandable spoken English.
My only regret is that I waited too long before beginning the recording and my voice had already lost much of its crispness and clarity — while miraculous, ModelTalker cannot overcome that poor start completely. But that’s another lesson — there’s no time like the present.
I hope my fellow PALS [people with ALS] will look closely at ModelTalker.com and start the process of recording your voice before it gets too weak to be effective.
I also hope that friends and families of PALS encourage their PAL to take these steps to preserve vocal capacity — an important element of the dignity and connectedness we all seek.
And finally, I hope that any foundations and government agencies seeing this will think about helping bring this technology to maturity and make it widely and easily available to people for whom it can make a significant difference.
This blog was originally posted April 25, 2013.
Editor's note: To learn more about the ModelTalker program, read Banking for the Future (voice banking technology helps people with ALS "capture" their true voices). In order to achieve optimum results with the ModelTalker, it is very important to bank your voice while it is still strong. Stuart Rakoff notes that he also has had good success with Speak it!, a text-to-speech app for the iPad that retails for $1.99 and offers several preprogrammed voices from which to choose.
About the Author
My name is Stuart Rakoff. I am 69 years old and live in Reston, Va. In the summer of 2012, I was diagnosed wtih ALS. There is nothing in my family or health history to explain why I have this disease. I started writing my blog, Drinking Through a Straw, as a way to help me articulate to myself a strategy for coping with my situation. As I shared these brief writings with family and friends, they urged me to share them more widely, so I began publishing them at Reston.Patch.com. I hope these writings will stimulate a discussion of the subjects I raise, and I encourage your comments.