When I was an undergraduate, pursuing my Bachelor’s degree in Speech, Language, and Hearing Sciences, one of the required courses was Speech Science. It was the mid-late 90s and we sat in a small “speech science lab” and shared large desktop computers with “high tech” software. We studied formants and looked at sounds and words on spectrograms and identified words by looking at speech sound waves. I have always been more of a clinical application person than a research scientist, and although I did well in the class, it was only because it was required. I didn’t love looking at speech in black and white representations.
When I asked how this information is used in the field, the professor told us that sometimes this type of visual feedback is helpful to people who are trying to produce target sounds. She also said that this type of technology is applicable to augmentative and alternative communication (AAC). At the time, I wasn’t interested in AAC either (which is unbelievable to me now!), but I remember her saying that speech scientists were trying to create more voice options so that women could use a woman’s voice and they’re also trying to create a child’s voice for children who use AAC. I thought, “Really? They don’t have that yet?”
Ten years later, when I walked into my first preschool autism classroom, it was the first time I had ever met anyone on the autism spectrum who was not able to talk. I watched four preschoolers with their corresponding adults in cubicles pointing to pictures and getting “tokens” every so often. The sound of Velcro, humming, vowel vocalizations, and occasional screaming filled the room.
I introduced myself to the teacher and she quickly said, “Good! You’re the one who handles this, right?” and she handed me a Dynavox MiniMo that was recommended for the one little guy. I looked at it with a mixture of trepidation and overwhelming responsibility as I responded, “Yes, it’s in my scope of practice.” I got permission to take it home and spent A LOT of time getting to know that little machine.
That first high-tech device I ever touched, did not have text to speech (synthesized speech) that is now built-in to the iPad. You had two options: (1) You could record everything onto the button (digitized speech) or (2) If the word was already pre-recorded in the bank of the device, you could link the sound file. When I wanted to add words that did not have a sound file, I would record the words with my voice on the spot, and write them on a list. One night a week (when the student had swimming), I would take it home and have my kindergarten son re-record the buttons in exchange for a later bedtime. It resulted in very cute developmental articulation errors (like “baftub” for “bathtub”) and the little boy having a unique voice that was a close match to his age. It was the best alternative I could find to avoid a mixture of sound files and an adult woman’s voice on a little boy’s AAC device.
As I became interested and immersed in AAC, I was excited to find out that there were women’s voices available and ONE child’s voice, which was really a higher-pitched variation of the adult voices, but better than nothing. BUT they weren’t on all of the devices at that time…only the supreme version of the high tech devices.
When we created Speak for Yourself and began to look for voice engines and text to speech options, there were no options that allowed for a child’s voice. We ended up choosing Cereproc, as many of you know and we selected the Caitlyn (Irish) voice as one of our options because we thought it sounded “young.” When Apple included text to speech in iOS 7, we incorporated the five additional English-speaking voices as options. The ability to change the rate and pitch for any of the voices is included in the app. We still email Cereproc about children’s voices periodically.
A few years ago, Acapela and AssistiveWare teamed to create the Josh and Ella child voices and we thought, “That’s more like it. They sound adorable.” About a year later, they made those voices available for license by other apps, but speech synthesis is expensive…to create, to incorporate developmentally into an app, and to license. Of course, as technology advances, prices and options become more accessible. We are always looking to provide the highest quality and most affordable choices so that people can make the decisions that are effective for the AAC users in their lives.
While we love the voices from Cereproc and Acapela, the issue of multiple AAC users in the same room having the same voice remained unsolved…until VocaliD.
When I first heard about VocaliD, it was through Rupal Patel’s TED talk . I watched with excitement. When it was over, I emailed it to Renee and our developer with a simple “I want this!” message, like a five-year-old with a Toys-R-Us catalog. We announced at the end of May that our companies will be working together to make VocaliD’s personalized voice service available on Speak For Yourself.
VocaliD’s cutting-edge technology blends an AAC user’s natural sounds with speech recordings from a matched donor to create a voice that is as understandable as the donor but reflects the individual’s vocal identity. Until June 23rd, 2015, Pre-order a VocaliD voice during their crowdfunding campaign at significantly below cost and be amongst the first few people ever to receive a custom crafted voice. If you want a voice but it is financially inaccessible, they recently a Need-A-Voice perk to help families who demonstrate financial need.
We are excited to be early supporters of this new technology, and we look forward to hearing all of the new custom voices. I wonder what has yet to be imagined in this area. As I’m writing, I’m wondering if people will look back at this in 10 years (because I’m sure everyone will be compulsively reading decade-old Speak for Yourself blogs) and think, “Ha ha…they thought that was ‘cutting-edge!’ There’s so much more we can do today!”
I sure hope that’s the case.
***Speak for Yourself does not benefit financially from any of the companies or services mentioned in this post.***
Leave a Reply