Speech recognition in Windows 10 & Office

Microsoft has quietly improved the speech recognition features in Windows 10 and in the Office programs. They’re still not great but you might want to give them a try if you haven’t talked to your computer in a while.

One fundamental truth has not changed: If you are serious about dictating on a computer, there is no alternative to the Dragon Naturally Speaking products.  The current version, Dragon Professional Individual 15, is top notch. (The previous version, Dragon 13, is still available, but it’s not as effective and seems to have some issues with Windows 10.)

Microsoft first built speech recognition into Windows more than 10 years ago, when Vista was released. It has languished ever since, appearing virtually unchanged in all the versions of Windows released since then. You can still find it in Windows 10 If you do a search for “windows speech recognition.” It’s clumsy to set up and not very accurate. There’s a reason you didn’t know it was there.

Last year Microsoft added a second speech recognition engine to Windows 10. There are no settings for it and it is used only for text entry, not for controlling programs or doing anything special. If your computer has a microphone . . .

•  Hang on. Let’s not assume anything. You know you need a microphone to dictate to your computer, right? If you have a laptop, it has a microphone built into it. If you have a desktop computer, you’re not ready to dictate until you get a headset with a microphone. They’re cheap. Logitech makes some good ones.

Where were we?

. . . If your Windows 10 computer has a microphone, hold down the Windows key and hit H. (You’ll have to memorize that keyboard shortcut – just remember it’s Windows key + H because “dictate” does not have an H in it.)

(That doesn’t really work, does it?)

(H?)

When you hit Windows key + H, a bar will appear across the top of the screen and the computer will be listening. You can start dictating.

Use the Windows key + H key to toggle dictation off and on. The microphone also turns off automatically if you start typing, or after a delay of a few seconds.

You can dictate anywhere that text can be typed – Word, Outlook emails, or any other program that accepts typing.

And the results are . . . well, they’re only fair, to be honest. I’ve been experimenting with different microphones on different computers. I don’t have any way to judge accuracy by percentages. I can only tell you subjectively that I have to fix more errors than I wish was necessary. When I’m dictating, I speak slowly without an accent in a quiet environment. Your experience might be far worse if you talk quickly or talk to your laptop in Starbucks.

With that in mind – try it! What do you have to lose? Some people are so comfortable dictating that this will be helpful even if it’s not perfect. It’s a big step forward from the previous Windows speech recognition.

Microsoft has basic instructions and a long list of the special commands that you can use with Windows 10 dictation – punctuation, commands to delete and select text, and a few more. Right now it only works if you use a US English keyboard; it may follow Cortana as it rolls out globally.


Office Dictate

Office 365 - Word dictate

If you have the latest and greatest Office programs, linked to an Office 365 subscription, then a new button may have appeared in the toolbars for Outlook, Word, Powerpoint, and OneNote. You can use it to dictate with a single click. Mary Jo Foley was told that the new dictation button is an outgrowth of “Dictate,” a project rolled out last year from Microsoft’s research labs.

I didn’t do anything to activate that button. It just appeared one day. If you don’t have a “Dictate” button, I don’t know what to tell you.

The Dictate button works similarly to the built-in Windows dictation that I described above. There is no way to tell whether they are identical, similar, or completely different under the hood. I did some experimenting, going back and forth, and did not see any significant differences – both were mostly accurate, both made mistakes, neither one seemed much better than the other.

The first time you click the Dictate button, you may see this window about “Intelligent Services.”

dictate_intelligentservices

“To provide these intelligent services, Microsoft needs to be able to collect your search terms and document content.”

If I understand correctly, when you turn on dictation with “intelligent services,” it sends your dictation online to Microsoft, where it can be analyzed by powerful cloud servers. It’s the same AI that Microsoft uses for commands addressed to Cortana. In theory, you’ll get far better results than if you depend on your local computer to do the best it can. But there’s a privacy aspect to it: you’re sending your spoken words to Microsoft and trusting them to treat them nicely and keep them secure and not sell them to advertisers. It’s roughly the same bargain that you make with Google when you talk to your Android phone or use Google Maps or Google Photos. Personally, I trust Microsoft, Google and Apple to be respectful of my privacy, but I live a simple and uninteresting life and you might feel differently.

Dictation is only one of a number of features that Microsoft is adding to the Office programs that use cloud services to do useful things. With your consent, you can highlight phrases in a Word document, for example, and get immediate translations to or from 60 different languages, look up definitions and Wikipedia articles, and get editing tips, among other things. Microsoft has an article here about the latest “Intelligent Services” for Office 365.


Google Docs

Google’s voice recognition has improved by leaps and bounds. Dictating text messages and searches on an Android phone has become second nature because it’s so accurate.

There is only one way to use Google’s server to handle voice recognition on a Windows computer. If you use Chrome, you can dictate in Google Docs. Your words will be transmitted to Google and the text will be sent back to your document.

To start, open a document in Google Docs with Chrome, then click Tools / Voice Typing. A microphone icon will appear; you can click on it to turn dictation on or off.

I did some experimenting with it and found that the accuracy didn’t seem to be any better than the Microsoft voice recognition. I was expecting a noticeable difference. Huh. Maybe more testing would reveal some differences. Your mileage may vary.

Go do some talking. In the next article, I’ll have something for you to listen to.

Share This