Voice recognition 1
Two students were originally involved, with a third beginning to use the system towards the end of the study. Of the two students originally chosen one was on a "statement" for general learning difficulties and the other was at Stage 3 of the Code of Practice. Both students were 15 years old and in the final year of their GCSE course. The third student was of the same age and also on a statement.
The main difficulty with voice recognition so far has been that the software only recognises a limited range of voices and is therefore relatively inflexible and inaccurate. The Via Voice package, in comparison, can be trained to recognise a particular voice to a very high level of accuracy. It also creates voice "profiles", allowing - theoretically - an unlimited number of users. This, coupled with its relative affordability (pound;399 including digital recorder) makes it an ideal option for schools, or for students who find the transfer of ideas to text a real challenge. The other feature much remarked upon is the idea of "continuous speech" - the computer doesn't need you to stop and let it catch up.
Going by the advertising, what you think you'll get out of Via Voice is a student with special needs effortlessly dictating essays. The end result, however, tends to be Kafkaesque due to recurring accuracy difficulties. The main problem, as usual, isn't the software but the user.
In order to train Via Voice, the student needs to dictate 476 relatively complex sentences. This is challenging enough or most students (and teachers), but unfortunately the problems don't stop there. Once initial training is completed the package needs several solid hours of dictation and correction to achieve its full potential, and the idea of concentrated practice with no immediate result is understandably frustrating for any teenager, never mind one with learning difficulties. Most students, especially those with special needs, find proofreading difficult and the "spellcheck" option is not able to assist them as Via Voice never misspells a word. Dictation itself also poses problems - the software may be able to manage continuous speech but most students (rather surprisingly) cannot.
All these difficulties are user-related, however, and can be overcome with time and the support of IBM and Olympus. Another problem is one of school resources - namely the amount of memory available on your computers. John Kelly Girls' Technology College is a well equipped school, but even it did not have the hardware available to run the Via Voice software. Most school systems are networked, and it's hard to find the 250MB of available hard disk space required. You also need 32MB of memory, Windows 95, a good quality 16-bit sound card and a double-speed or faster CD-Rom. Coupled with a processor requirement equivalent to Intel Pentium 166mhz with MMX or faster, this appears to put the software out of most schools' hardware league.
It would be a wise move by suppliers to offer a computer, digital recorder and Via Voice software as a complete package if they wish to target the school market.
Despite the problems encountered, voice recognition technology has now become good enough for it to be a useful addition to a schools technology department. Via Voice lets you control a PC by voice, a feature which could revolutionalise the life of any student with motor co-ordination difficulties, and particular forms of dyslexia would benefit from the "playback" option, where the computer reads back what you have written. Those in accelerated learning programs would be able to produce a great deal of text using this software, and it has the potential to take over the time consuming job of scribing for students with a range of needs. The beauty of the Via Voice system is that only one needs to be purchased by the school and can then - especially in conjunction with the digital voice recorder - be used by as many students as necessary.
IBM's Via Voice speech enquiry lineTel: 01475 555 047www.ibm.comviavoiceuk