Welcome to the Aidis Trust blog. Here you’ll find our posts on assistive technology that are meant to inform and encourage discussion. Feel free to join in!

Giving Technology A Voice (part 1) – Voice Banking

P2G2

 In my 14 years of working with disabled people and the technology designed to help them, I have spent a lot of time with Augmentative and Alternative Communication (AAC). The type of AAC that I have been involved in is provided by computer technology, a branch of electronic assistive technology (eAT).

A brief history

From my own personal experience, in 2000, AAC devices were all made bespoke. Built specifically for disabled people who, for a variety of reasons, couldn’t communicate verbally. These devices were mainly priced around the £6,000 to £7,000 price mark and all did a pretty reliable job of providing a synthetic voice for those without their own.

The voices of the past

Synthetic voices at the turn of the century didn’t fool anyone into thinking they were anything but computer generated. Stephen Hawking’s voice is a classic example of the standard voices that were available.

It wasn’t too long before the quality of voices started to come through. The first decent quality voice I remember well were the voices by Acapela and these started to be used in AAC devices. Around the same time, ScanSoft (a company that at the time owned the Dragon Dictate speech recognition licence) had produced probably my favourite voices called RealSpeak. There are other speech engines such as the SAPI voices, commonly used by Windows for its Narrator screen reading and Ivona, who have developed a whopping 50 odd voices, including Welsh.

Nowadays, we have synthesised voices (called Text-To-Speech or TTS for short) in many different languages and indeed accents. RealSpeak voices are currently available in 16 different English accents – English, American, Indian, Scottish and Irish!

Here’s an example of how expression can now be built into TTS voices –

 

 

TTS technology is being used in many places today, from the train announcers at railway stations to the automated call handling – ‘if you want technical support, press 1, press 2 to talk to a human, press 3….’ And so on.

Bespoke systems are now also being joined by ‘off the shelf’ solutions, offered by the range of Windows, Android and Apple’s iOS tablets.

Saving your voice for a rainy day

Using a synthesised TTS voice is fine, if you’ve never had any voice. But what about those people who have had a voice but through an accident of disability, have lost the voice they’ve use all their life?

The answer – Voice Banking

Voice Banking enables people to keep a copy of their voice, should they lose their voice. Such scenarios could include stroke survivors and motor neurone disease.

The science behind the technology TTS works by breaking up words into their constituent syllables, to be pieced together again, to make different words. An example that Acapela uses is the word ‘impressive’. Impressive is then using ‘im’ from ‘impossible’, ‘press’ form ‘president’ and ‘ive’ from ‘detective’. So, if a computer has learnt how you say the sentence ‘it was impossible for the detective to find the president’ then other words such as ‘impressive’ could be cut and pasted together.

We are about to trial a service by an organisation called ModelTalker, to see if we can have our own TTS file built. Apparently, this will involve recording 1,600 phrases, which should enable them to recognise the way I speak every syllable, we shall see. As soon as I have the results, I’ll post examples on the blog for you to hear.

If you’d like to read some of our views and reviews on a selection of AAC products, please visit http://www.aidis.org/category/speech-and-language.html

Share Button

Tags: ,

Trackback from your site.

Leave a comment