Dragon speech recognition software was first released in 1997 by Dragon Systems of Massachusetts. The company was founded in 1982 by Dr James Baker, (who had described the principles of a speech recognition system in 1975, releasing Dragon Dictate for DOS). At this time however, the computer hardware was not powerful enough and the software required that users pronounced one word at a time, with a small pause between each, because it could not distinguish the boundary between one word and another during continuous speech.
Dragon Systems was purchased by Lernout & Hauspie, a Belgian company, in 2000, but they were declared bankrupt in November of that year. At that time, it appeared that the software might cease development, but the rights were acquired by ScanSoft in 2005, who also acquired Nuance Communications, and rebranded themselves to Nuance.
The early versions of Dragon used this discrete speech and the results achieved were, to be honest, disappointing and frustrating. When I first started using the software myself, and training people in how to use it, I was working with an organisation called the Foundation for Communication for the Disabled (now AbilityNet), and both the software and the computers systems being used were very expensive, and although many people could see the potential, it was not good enough, bearing in mind that many people using it were simply unable to use a mouse or a keyboard. It also required a great deal of investment in time to ‘train’ the software to recognise your voice, by reading aloud long lists of words and commands as they were displayed on the screen.
A competitor was available with an IBM product called ViaVoice but development did not match that of Dragon, and the voice commands available were a bit limited when compared with Dragon. In my view, the accuracy was not as good either, and the product is no longer supported.
Gradual improvements in the technology began from the release of version 5 and now, people were able to speak continuously. My own feeling is that from version 9 and 10, the software was beginning to become more usable, so that not only could it be used at home, but also in the office environment. Much of the training that I completed at this time was as a result of assessments for the Access to Work scheme, a government grant still available to help provide some support and equipment for people with a disability in the workplace.
Even now though, if you speak to people about their own experience with the software, they would probably describe it as disappointing, and in the office, the IT support teams would almost certainly say that it did not work!
The software itself was gradually improving so that the speech engine, the component that recognises what is said, was becoming more accurate, the hardware was becoming more powerful (and cheaper!), and the quality of the microphones available to use was also improving, so that Nuance claims a very high accuracy level for the latest version, 12.
Windows also has a free version of speech recognition, and it can work for some people, but the list of commands available is a bit limited, and I do not believe it is as powerful as Dragon. It is also quite frustrating to use, whereas Dragon is more natural.
Nuance acquired the rights to a speech recognition program for MAC called Macspeech several years ago, and rebranded it as Dragon Dictate for MAC, currently available as version 4. It is not yet as flexible as the PC version for those people who need to be able to work hands free, but the accuracy is very good, and development is happening very quickly, so this will improve with each successive release!
