Speech Recognition Software

Pros and Cons of Speech Recognition Software

1. Good points

• You speak and your words are written on the computer screen.
• All the words that appear on the screen are correctly spelt.
• You need fewer keyboard skills.
• You can say word-processing commands,
e.g.. 'New line', 'open brackets', 'Open file', 'Page down'.
• It saves time. Speak a couple of chosen words, e.g. “My address” and it will insert your whole address.
• Some systems allow you to playback the words you have said.
• This helps with poor short-term memory and proof-reading.
• This can be very beneficial for dyslexic users.
• The software price is from £50 upwards to several hundred pounds.
• Dyslexic pupils have used SR in exams.
• It can be used in word-processors, E-mail, spreadsheets and practically anywhere, and is most effective in the system's own VoicePad.
• Some dyslexic users have found it liberating.
• Many comment on improvements in spelling and reading as they always see correctly spelled words.
• It is particularly helpful for students, young adults and adult users and those familiar with computers and word processing programs.

2. However

• You need a powerful, modern computer for the newest and best SR systems.
• A good soundcard is essential, or a laptop that you know is good with speech input.
• SR generally works better on desktop computers than on laptops.
• The microphone must be of reasonable quality.
• You must check the position of the microphone each time.
• If you have problems, the first thing to do is to readjust the audio settings.
• You have to learn, and then remember, how to use the software and say all the commands and punctuation consistently, although the latest software will have a go at putting in commas and full stops for you.
• The software has to learn your voice and cannot be trained by someone else for you.
• You need dictation skills. Oral composition is not always easy for everyone. It often helps if you plan, draft, and use concept mapping programs like Inspiration and MindGenius to prepare your work.
• You need to be able to dictate in the written language, which is different from how we normally speak.
• Proof-reading is hard, especially for dyslexic users. Can you spot correctly spelt words in the wrong places or where the words you spoke have been missed out?
• Correcting errors is time-consuming, so accuracy is important, to reduce errors.
• The system recognises words in context more easily than single words. When correcting, it is better to repeat a short phrase than a single word. 
• So it’s much better to speak clearly, with a little more stress than usual on the little words (and, the, a, any, to), than to try to speak fast.
• You must continue to train the system so that it improves.
• A success rate of 90-98% is claimed, i.e. one word in ten may be wrong. This may be better than dyslexic users achieve with typing or writing.
• People with accents or speech impediments may need more practice.
• Boys may need to retrain the program after their voices have broken.

3. Systems Needed

• There are SR systems from Dragon and IBM, each with several versions.
• It is rarely necessary for a dyslexic person to buy the most expensive.
• Each version is usually better than the last, although the older one may still be available more cheaply.

Conclusion

Speech Recognition is amazingly clever. It is an important way forward for the future for persevering secondary age and adult dyslexic users rather than younger ones.

The technology changes regularly so that we suggest you check current information.

Where to find out more:

iANSYST stocks all the systems. The web has a comparison of the programs.

Page updated: 27/02/06