There are four main factors you should consider when choosing speech recognition software:

• usability - how easy is it to learn how to use?
• accuracy - how much of your dictation does it get right?
• reliability - how often does the program crash?
• capabilities - how extensive is the command set for advanced users? That is, what can you do with the software?
• compatibility - can you compose a letter in Microsoft Word? What about spreadsheets? Can you email?

Usability

Setting up the software involves installing it and creating profiles that the software can use to differentiate between the speaking styles of different users. Although they are "perform once" tasks, you still need to consider how easy it is to complete them.

Then there is the day to day use of the software. Discrete speech requires you to pause between every word. It makes it easy for the program to interpret what you say, but slows down transcription and makes communication with the program unwieldy. Continuous speech is at the other end of the scale and characterises the way we talk to our friends and colleagues. Because spacing between words is minimised, continuous speech is harder for speech recognition software to understand. However, it is much easier for use to speak this way, and of course it makes for speedier transcription.

Your speech recognition software should be able to cope with continuous speech for stress free use.

Accuracy

There is no point in buying software that is cheap, easy to "train" and quick to operate if it keeps getting your words wrong. Accuracy measures the proportion of correct interpretations your speech recognition software makes. Nuance claim that Dragon Naturally Speaking 10 achieves up to 99% accuracy. Higher levels of accuracy mean less time spent making corrections and "teaching" your software your intended meaning.

Reliability

Reliability of software is less of an issue for speech recognition software these days, however, there are sometimes programs that develop a reputation for being unstable. Do your research.

Capabilities

Do you need the program to do anything of a complex nature? Most users use only a small subset of a program's functions, but your needs may be different. Check the technical description of the software to make sure it can do what you need it to do. Also, don't forget basic requirements like selecting text without using the keyboard. Can the software do that? What about functions that are not related to dictation and transcription, like the ability to control your PC?

It's possible that other people may need to use the software you buy. Can it cope with multiple users?

Compatibility

Most people use or have used a number of other programs such as Microsoft Word, Excel and Outlook. These are very popular programs that your speech recognition software should be able to work with. Thankfully, most do but you should check nonetheless.

Tips For Buying Speech Recognition Software

As with any other product you would buy, you are advised to do lots of research. If you get an unsuitable program, it's not just the money that you have wasted. Think of the time you have spent waiting for it to arrive, installing and setting it up and training it to recognise your voice. You can do a certain amount of research in places like Amazon. As well as technical descriptions offered by the manufacturers, previous customers who have bought and actually used the product often write reviews which can help guide your purchasing decision. Remember - the more reviews, the better informed your decision will be.

Most providers offer their software on a trial basis that lasts a certain period of time. This is very useful for allowing you to get familiar enough with the software that you know whether it meets your requirements.

Find out why so many people choose Dragon Naturally Speaking 10.