It is a strange sight to watch anyone talk to a phone in the middle of a lunch hour at a South Indian restaurant. Before you raise your eyebrows, I said ‘talk to’ a phone and not ‘into it’ like we all do. But then there has been a considerable uptake in the number of people saying “OK Google Now”, thanks to the success of Moto phones in the past few months. However, voice control is no longer a feature that is limited to a Motorola smartphone.
In fact, voice control captured the attention of smartphone users when Apple launched its Siri voice in the iPhone 4S in late 2011. So people all over the world were getting Siri to do stuff for them – all the way from skipping songs to booking tables at the snooty downtown restaurant. But Siri was a dud in countries like India, where we speak English very differently from those who trained Siri. Those of us who had spent a couple of days in the West could fool Siri, but not the entire iPhone janta.
This has been the biggest problem with voice control. Even Microsoft Windows has had voice recognition for a few years. It works, provided you had spent considerable time training the machine to understand the way you speak. So you know why it is hard to get it right when millions of people are blurting out voice commands all over the world.
On the flip side, it is this increase in number of users that is helping this technology become better with every passing day. With the success of voice control based primarily on the ability of the machine to match the command with the data base, a larger, ever-increasing database can only make life better. And the biggest example of this is Google voice search.
Google has had voice search for sometime now. Most of us have tried it before and realised that Google has no clue what we are trying to say. But things have changed. Just last week Google India became confident enough to announce that its voice search was good enough to recognise Indian accents. Sandeep Menon, Google India’s head of marketing, said they achieved this by deploying 700 volunteers across India to collect voice samples, which were then analysed by their engines. Interestingly, he asked more Indians to use the service, which is based in the cloud, so that