facebook rss twitter

The latest Chrome browser includes speech recognition API

by Mark Tyson on 26 February 2013, 13:00

Tags: Chrome

Quick Link: HEXUS.net/qabs7b

Add to My Vault: x

Google has updated the stable release of the Chrome browser with support for the Web Speech API. This API will be useful for web developers wishing to add speech recognition capabilities to their web apps. Another new audio feature, present in early builds of the Chrome browser in testing, shows users which tabs have sound playing within them.

Web Speech API

The latest stable version of the Google Chrome browser includes speech recognition capabilities. Google announced it via the Chrome blog late last week “Web apps, listen up. Today's Chrome Stable release includes support for the Web Speech API (discussed last month), which developers can use to integrate speech recognition capabilities into their web apps. To see what this might look like, check out this demo on how to compose an email just using your voice.”

Voice control and speech recognition often sounds a lot better in theory than it works in practice. My own little test of the feature showed that it doesn’t even know its own name... I repeatedly said “I am testing the Web Speech API – comma – it doesn’t work very well.” It always had problems with the word “speech”, wanting to use the word “page” instead, though sometimes it got “API” correct.

“I am testing the Web Speech API – comma – it doesn’t work very well.”

The latest Chrome 25 stable release also disables silent extension installs in Chrome for Windows, this ensures you give consent to every extension installed and “keeps Chrome fast and safe”.

Noisy tab indicators

Google is working on audio notification icons on Chrome browser tab-tops which are playing or recording audio. These alerts could be useful for a number of reasons:

  • To help users hunt down annoying streaming adverts among their many open tabs.

Or conversely

  • To keep open music streaming tabs you don’t want to close by accident, or to exclude these open tabs from those Chrome automatically discards when memory runs out.

TNW reports that the audio tab indicators are in the “Canary build” of Google Chrome and the notification appears if sound is playing via a variety of methods including Flash or HTML5. The notification is a little animated bar type “VU meter”.



HEXUS Forums :: 4 Comments

Login with Forum Account

Don't have an account? Register today!
I love the little features like the sound icon, but talking to my browser whilst it sounds great is probably just one of those “ooh shiny” features nobody ever uses. I've tried the Android voice input feature (and Google Now) and it works a charm, apparently my mildly posh Queen's English voice is well suited… But, I do 99% of texts, emails or other input in public places and I don't want the world to know what I'm texting the mrs! Same at home, does she really want me sitting on the sofa chatting to Chrome whilst she's watching Mr Selfridge… I'll end up visiting “www dot hexus dot will you shut up I'm watching this”.
The only place I can see me using voice recognition is in a car.
kingpotnoodle
I do 99% of texts, emails or other input in public places and I don't want the world to know what I'm texting the mrs! Same at home, does she really want me sitting on the sofa chatting to Chrome whilst she's watching Mr Selfridge… I'll end up visiting “www dot hexus dot will you shut up I'm watching this”.
That last bit made me smile - nice to know I'm not the only one… ;)

Speech dictation leaves me a bit cold - yes I know it works pretty well these days, but to my mind you still need a quiet location. Useless in the open plan cubes that predominate. On the other hand speech command strikes me as useful - the command set is limited and should be possible to get with high accuracy. What I have issues with is that surely with Chrome you'll be doing predominantly dictation - and dictation of non-words too. So any system that's expecting letter-style English is going to have issues with “www dot hexus dot net”, and I suspect you'll spend a long time spelling out stuff like xkcd, and stopping it putting in a space in “theregister”.

The “noisy tabs” stuff sounds like a good idea - but it's only going to achieve full usefulness if you can not only identify that loud advert, but also zap it. Not something - based on current actions - that Google will want to let users do.
“Google Search: House prices in putney”

Now Searching for… Hardcore Pornography

“Incorrect”

Input invalidated, please state command clearly

“Google Search: House prices in putney”

Now Searching for… Hardcore Pornography

“Incorrect”

Input invalidated, please remember voice is trained by your previous inputs

“F*** you”

I've been trying to help you with that


The possabilities of a turing bot as a search helper and this API are endless :D