WebWise news report - Voice recognition software
Voice recognition software - an assistant with selective hearing
No longer just a cool add-on, voice recognition apps are once again in the news, but this time as potential personal assistants for phone users.
I first became both aware and in awe of speech recognition technology about 15 years ago when my computer-whizz brother attempted to install what we thought would be revolutionary software on our home PC. For me the idea that something without a brain could understand language was ludicrous, yet as a then rather slow typist, I was equally excited at the prospect of bypassing the keyboard altogether and simply instructing my computer to write my schoolwork. I was however bitterly disappointed - and I wasn't the only one.
As ³ÉÈËÂÛ̳ technology guru Rory Cellan-Jones explains, the legal and medical industries had high hopes for speech recognition, with busy professionals expecting to dictate letters to their computers, rather than their secretaries. The reality was however that their assistants were still needed to correct the sentences that had been lost in the process.
Smartphone giants Apple and Google have both featured voice recognition software for several years, but they're now hoping to demonstrate how it can move beyond a party trick, to a feature that can be fully integrated into daily life.
If you pine for days gone by of servitude and serfdom, and just want to tell something what to do, this may be the app for you. When I command my app to "Search ³ÉÈËÂÛ̳ WebWise" for example, I'm taken to the correct search results page. "Navigate to London Bridge" is also recognised perfectly. But as a timesaving app for say, sending a text, making sure I speak slowly in my best RP voice isn't exactly quicker than pressing my messages icon.
The things that take the most time tend to require more thought, and knowing exactly what you want to say in an email without seeing the words on the screen is no easy task. Saying things like 'OK' or 'See you there' is straightforward enough, but consider the effect of different accents, place names and surnames, not to mention changing your mind about what's just been written, and suddenly the perfect time-saving speech recognition app seems a very long way off.
Here are just some of the results I got when I tried to say "³ÉÈËÂÛ̳ WebWise is a beginner's guide to the internet" in a voice recognition email. Some of the results were too rude to publish:
- ³ÉÈËÂÛ̳ website is again is god's to be on tonight.
- ³ÉÈËÂÛ̳ web, why is it again is sky to the internet.
- Baby see but why is a the game tired to the internet!
Read about Charlie Swinbourne's week spent monitoring speech-generated subtitles.
Comment number 1.
At 5th Jan 2012, voiceguy wrote:You right. About 15 years ago was when I got into speech recognition. It didn't work that well due to immature technology, slow PCs and poor microphone / soundcard quality. is probably the software you are referring to.
Dragons come a long way now though. Corei7 PC, 8 gig ram and SSD really make dependent voice recognition fly - and its pretty accurate. I made a few demos of the latest version and put them on youtube of dragon 11.5 if you are kene to see how it runs on a decent PC. It works on Mac now too.
Independent speech recognition - Siri - Dragon Dictate for Iphone and the android versions are developing in leaps and bounds. For now though you can get some pretty amusing results, as you pointed out.
I think in the next 3 years in independent speech recognition for mobile devices will drive the majority of speech recognition market growth.
cheers
Complain about this comment (Comment number 1)
Comment number 2.
At 10th Jan 2012, Olivia Olson wrote:This comment was removed because the moderators found it broke the house rules. Explain.
Complain about this comment (Comment number 2)