Wednesday, September 06, 2006

My Voice Project

Hurray! I'm happy to say I've finished my 'voice capture' project. It took me a little over a month to complete. I don't know if you know it but my voice is just about gone. I went from talking like a drunken sailor three months ago to now like a man with rocks in his mouth. I'm really pleased at this point with me synthesized voice.

I can thank Dianne Lyons, my speech therapist for linking me up with Dr. Tim Bunnell. He helped me capture my voice in a speech databank. Tim is the Director of the Center for Pediatric Auditory and Speech Sciences (CPASS) and the Speech Research Laboratory at the Alfred I. duPont Hospital for Children in Wilmington, DE. The focus of most of his research is hearing and speech problems in children, but one of his projects led to the development of the ModelTalker system which has been of great interest to people with ALS. Checkout this site:

The picture above is a screen shot of the computer application that I used along with my computer microphone to record my voice. The requirement was to repeat 1649 utterances. If you look closely, you can see an example of utterance #670 of 1649. Some are simple utterances like 'Hello' or 'I am'. Others are meaningless to me but important to the speech analyzer like 'his ard is arge'. There are even funny ones like 'my nerd farm'. It usually took me a few minutes to recover from laughing so hard after being prompted to repeat such a ridiculous phrase. Also, notice the three measurement gauges. Pitch and Loudness didn't give me problems. But Pronunciation was my biggest challenge. Most of my attempts were in the red zone. But after repeated attempts, I managed to score around 70%. I was constantly challenging myself to get the best pronunciation score as possible.

When complete, the next step in the process was to upload the 1649 utterances to the ModelTalker server. My voice was analyzed and transformed into my new synthesized voice. I then downloaded the final synthesized voice files to my computer. I have a few applications on my laptop that allow me to type words and phrases and instead of getting a fake computer-like-voice, people hear my own voice.

I have to admit my voice is not perfect because I started too late. So my message to my fellow PALS is:

A significant subset of ALS patients lose the ability to speak relatively early on in the progression of this disease and for those people, ModelTalker may provide the ability to 'bank' their voice while they are still able to speak and then use a synthetic version of their own voice in a communication device when that is later needed. Did I mention it's free.

Go to Dr. Bunnell's site at: to get started! Tim is an expert in his field and a nice guy to work with.

Here are two sites that provide communication aids for the disabled:

E-triloquisttm is a FREE PC-based communication aid for a speech impaired person. It serves as an electronic voice for those who can't speak on their own.

EZ Keys XP allows the literate user to do everything from typing a letter, to engaging in conversation with a friend, to exploring the worldwide web. World-renowned astrophysicist Stephen Hawking uses the software to communicate, write papers, and deliver lectures around the world.