I mentioned before the good service provided by SpinVox when I had a few problems getting the service to work initially. Well, the other day I had another problem with the SpinVox service: callers were experiencing a 10 second period of silence before hearing the SpinVox voicemail greeting. A quick call to customer service, during which I was able to speak to a real person, soon solved the problem.

I’ve finally given in and purchased Dragon NaturallySpeaking preferred version 9.5. I only needed an upgrade because I have version 8; however, it’s cheaper to buy the full version from PC World or Amazon than to buy the upgrade from Nuance. It was the little things like apostrophes, capitalisation and dictating brackets that got to me in the end.

NaturallySpeaking is so much better than Windows Speech Recognition. I’ve started from scratch with a new speech profile, hardly giving it any training, and already it’s much more accurate than Windows Speech Recognition ever was. As far Windows Speech Recognition Macros is concerned, I’ve always had this facility in Dragon NaturallySpeaking with Vocola and NatSpeak, and now I have them running under Vista.

I’m still a bit peeved that I had to pay the price of the full version instead of getting a decent discount for buying an upgrade. Just to be clear, the full version costs £145 from Nuance or £100 from PC World or Amazon, and the upgrade from version 8 to version 9 costs £100 from Nuance.

I’m being more and more tempted to upgrade to Dragon Naturallyspeaking Preferred version 9 for speech recognition on my Vista computer. I know, I was initially very impressed with Windows Speech Recognition, but the more I use it, the more frustrated I get.

There are one or two niggles, which I’ll come to shortly, but the most annoying thing is that using the correction dialogue and navigating about a document by voice has become very slow. When I issued the correct command, the alternates panel used to come up reasonably quickly; now it can take as long as half a minute. And it can take up to a minute after I’ve chosen the correct alternative and said OK. Something seems to have happened to slow these functions down dramatically. I can’t find any mention of anyone else having had this problem, but there’s not much discussion on the Internet of Windows Speech Recognition anyway; and that’s also a problem: Dragon Naturallyspeaking has a very active support community.

I’ve uninstalled Windows Speech Recognition Macros and deleted the English United States voice profile that I had created in order to see if Macros would work and I’m now testing to see if navigation and correction have speeded up.

The other niggles I have are:

• How do you dictate an apostrophe followed by the letter S? In Naturallyspeaking you simply say “apostrophe S”. This doesn’t work with Windows; you’ve either just got to say the word normally and then correct it or you have to resort to telling the computer the individual keys to press.
• How do you set up user defined shortcuts? For example, in Naturallyspeaking if I often use the phrase Microsoft Corp. Ltd., I can enter this phrase into the speech dictionary and specify that I want it to be typed whenever I say, “MC Limited”. I discovered that you can actually do this in Windows Speech Recognition, but it’s not immediately obvious how to do it and you need to remember the phrase that you will pronounce as it doesn’t appear in the user interface of the speech dictionary.
• Capitalization is a bit of a pain. You can’t tell the recogniser to capitalise every word from now on until you tell it to stop. You have to either precede each word with “Caps” or, once you’ve finished dictating the phrase, tell it to capitalise the previous so many words.
• As is dictating brackets. Instead of saying something simple like “open bracket” you have to say “open parenthesis”.

On the bright side, since uninstalling Windows Speech Recognition Macros and deleting the voice profile that I’d set up using the English United States language setting, navigation and correction have returned to their former, acceptable speed. Perhaps it’s not a trip to PC World today after all (yes, they, too, are selling the full version of Dragon Naturallyspeaking for around £100, i.e., the same price as Nuance will rip you off for the upgrade).

This is a service that just works the way it says it does. You sign up for the service, follow the instructions to divert your unanswered mobile phone calls, and after that you get your voicemail messages sent to you by text or email. I usually get an email of a voicemail on my BlackBerry within no more than 15 minutes.

Unlike other services (such as BT RingCentral) SpinVox’s support is accessible and responsive. In setting up SpinVox I had a slight problem in that it wouldn’t use my personal greeting. I phoned SpinVox’s support and explained the problem. They said they would phone me back within an hour with a solution, and they did.

I tried out a new telephone service from BT called BT RingCentral. You get a 0800 or 0844 number and you can set it up with voice menus to direct people to the appropriate extension. The extensions can be landline or mobile numbers. It’s supposed to make your little company sound like a big company to callers. Sounded like a really good idea.

Unfortunately, it doesn’t quite work, and you get all the poor service that I’ve come to expect over the years from BT. Here are just a few of the problems I had:

• It’s supposed to transfer the caller ID to the receiving telephone that you’ve set up as an extension. In my case this was a standard BT business line. Transferred calls always showed as Out of Area. The RingCentral call log showed the correct caller ID and I have no problems with caller ID on the BT business line with any other direct or transferred calls. RingCentral seemed incapable of transferring the caller ID.
• The number you get from RingCentral is supposed to be able to answer voice and fax calls. I found that if the receiving telephone has an answering machine then RingCentral starts sending what sounds like fax tones instead of putting the caller through.
• Several times while testing the system as if I was a caller phoning our business I got a recorded American voice stating that the RingCentral number had not been activated. I mean no offence to people from America, but that would create a good impression to our customers, wouldn’t it?
• You’re supposed to be able to transfer calls that you’ve received on one extension to another extension, where the other extension is, again, another telephone number. This is supposed to be as simple as pressing * and then the number you’ve associated with the extension. I did get it to ring the extension, and the caller got music on hold. You’re supposed to then simply hang up and the call gets transferred to the other extension. This never happened: the caller was left indefinitely listening to music on hold.

I raised several support tickets over a period of three days with BT RingCentral and got no response. Eventually I canceled my trial. After that I have received emails and phone calls from BT RingCentral saying that they want to help sort out the problems. Unfortunately, it’s too late as far as I’m concerned.

I’ve found out why Windows Speech Recognition Macros doesn’t work on my system. In the Windows Speech Recognition settings I have the language set to English United Kingdom. If I change this to English United States then the macros that I’ve created work; however, speech recognition is, obviously, not very accurate because it doesn’t like my Scottish accent.

Microsoft has made available an add-on to Vista’s Windows Speech Recognition (Windows Speech Recognition Macros) that allows you to set up your own voice commands that will be performed when you say specified phrases. I was quite excited about this because I was missing this facility from Dragon Naturallyspeaking. Unfortunately, I can’t get it to work. I can create and save macros, but when I say the phrase that’s supposed to run the macro all it does is type the phrase literally.

Never mind, thanks to another blog (Rob’s Rhapsody), I found a way of doing what I want using shell scripts. I can get any script file to run by simply saying, “Start…” followed by the name of the .vbs file.

This really is excellent! Now I’ve got all the features I used in Naturallyspeaking without having to pay the £100 for the upgrade to a Vista compatible version. I don’t mind having to pay for an upgrade, but the price Nuance is charging for the upgrade is the same as Amazon charges for the full version. I feel Dragon users who have upgraded for many years should get a better deal than that. Now that such excellent speech recognition software is available for no extra charge in Vista there’s no need for new users of speech recognition to pay for third-party products.

I’ve been using speech recognition for over 10 years, starting first with IBM’s Viavoice and then through several versions of Dragon Naturallyspeaking. The PC I mainly use now has Vista and I was very reluctant to pay for yet another Dragon update; version 8 won’t run under Vista and it would cost about £100 for the upgrade. So I decided to try the speech recognition facilities included in Vista.

It’s not quite as accurate as Dragon Naturallyspeaking, but that’s only to be expected as I’ve invested years in training the Naturallyspeaking speech files. In the few days I’ve been using Windows Speech Recognition, though, its accuracy has improved considerably. If it continues to improve then I’m certain I won’t need to fork out for the Dragon update.