Sunday, 23 August 2015

How Use(less) is Voice Recognition on a Mobile?

Speech recognition has been around for decades. Heck, my first IBM compatible PC ran on Windows 3.11 for Workgroups and I installed some speech recognition software from Willow Pond Audio.
Combined with an external microphone that I plugged into the PC, I could set up voice commands such as "Computer. Open Microsoft Works".
It was hit and miss, slightly slow, but it did work.

Twenty years later and my first proper PC is massively underpowered compared to even the most basic mobile phones, so has voice recognition improved in the last two decades?

For many people, voice search (or voice recognition) is now part and parcel of their smartphone's operating system. Less than ten years ago, with the dawn of Bluetooth connectivity, voice dialling and recognition became useful, and for the most part, it worked.

In 2015, things have improved slightly, but we are still a very long way off being able to properly control our mobiles and connected devices by using voice recognition.

Apple's Siri stands out as possibly the most widely recognised voice search program on a mobile phone, mainly due to the massive uptake of the iPhone. Does Siri do a great job, or is "she" more of a novelty?
Google's own Google Now is a clunky affair, made too big for it's own boots through attempting to connect every aspect of our lives from plane tickets, online orders and telling us how far we've walked in the last month. Saying "Ok, Google" can be a lot faster than swiping/typing away on the on-screen keyboard, but let's face it - you look like a plonker talking to your phone when it's clear that you aren't making a call.
Microsoft's version, Cortana, has emulated a more "personable" response in "her" replies, but "she" is still very hit and miss.

Just before Google Now launched, I tried Samsung's S-Voice on my Galaxy S3.
The first time that I used it I was amazed (or was it more of a novelty?). However, the first time that I needed to use it in anger whilst I was in my car, it failed miserably. Repeatedly.

Google's voice search does work far better, and disabling S-Voice was a godsend, but all of these methods rely upon having a decent internet connection.
Many people (at least in the UK) do have good 3G connectivity, though I hear many stories of how even in large cities maintaining a fickle stable internet connection can be infuriating. Not being able to use hands-free on your phone due to this lack of internet connection renders the technology worthless.

Only two weeks ago, I was driving to work through the beautiful Hampshire countryside when I needed to make a hands-free call. The smartphone that I had to hand was a Microsoft Lumia 535, so I thought that I could rely on Cortana to help me to call a colleague at work.
Cortana dutifully obliged me by opening and sat in my mobile's cradle, awaiting my bidding.
Two attempts to get her to call a colleague failed. At first she couldn't understand what I was trying to say - This is never a good start for voice recognition.
Then, Cortana opened my Met Office app for me. Once again, not at all what I wanted.
Being rural Hampshire, my tenuous link to the internet was lost, and then Cortana revelled in telling me that "The Internet and I aren't speaking at the moment".
I gave her two more chances before giving up and waiting until I saw my colleague at work.

Hello! Cortana?? I'm on the phone!!!


I just don't get it. Why should a smartphone packed with all of the latest technology, processors and RAM be totally unable to undertake simple tasks without a data connection?
Ten years ago, there were plenty of phones with voice dialling. You could just ask them to call a contact, and the number would pop up and could then be dialled.
Smartphones really aren't that smart if this basic functionality is lacking.

I did have a slight dig at Microsoft on Twitter a few days ago, when they posted a Tweet about how Cortana could help you learn something new every day. My reply was " it's a pity that can't learn anything without being connected to the internet.
She can't even make a phone call without data", which sums up my feelings about voice recognition in general.