Voice recognition: is it time for web managers to listen?

Technology that deciphers human speech is fast gaining traction, with Apple, Amazon and Microsoft leading the charge, Scott Payton says. What should web managers be doing now to prepare?

It feels like the future has been arriving particularly quickly this year.

The world's first self-driving taxi trial began in Singapore in August.

Sales of 'wearable' and 'smart home' devices like internet-enabled watches, lights and thermostats shot up.

And a slough of (relatively) affordable virtual reality headsets arrived in shops.

But another technology fast gaining traction is more likely to shape the future of corporate websites and, perhaps, transform the way people use them: voice recognition.

Speech-to-text software has been around for years. New Jersey-based Bell Laboratories created the first system, named Audrey, in 1952. Google created its first voice search tool in 2002, and started putting it in apps in 2009. 

The difference today is that such technology is starting to work really well.

I bought an Amazon Echo last weekend – a tube-shaped speech-controlled speaker. It's extraordinary. I reckon that it understands 95% of what I ask – from turning on the dining room light, via changing the radio station, to what £1,000 is in euros (not much).

Like other ‘smart’ devices, the Echo is no longer a just nerdy niche: Amazon is estimated to have sold more than 5 million of them in the US alone.

Alexa, the voice recognition software that powers Amazon’s Echo (as well as a range of the firm’s other devices), is getting cleverer all the time. Amazon has more than 1,000 people working on the technology – which also improves itself by learning from users’ past speech requests.

Siri, Apple's voice assistant, is getting smarter and more ubiquitous too. Importantly for corporate web managers, Siri was installed this year on millions of Apple laptop and desktop computers.

The same is happening for PC users: Microsoft's voice assistant 'Cortana' arrived on desktops via Windows 10 last year.

Voice search is also built into Google's Chrome web browser, as well as various Google apps.

And Microsoft announced in October 2016 that it has developed speech recognition software that has reached 'human parity'.

Why is all this important for corporate websites?

A notable trend this year has been for companies to put big natural language search boxes onto their home pages. Visit GE.com, Eni.com or TNO.com for examples.

General Electric's home page is among a growing number that encourage visitors to use natural language search

General Electric's home page is among a growing number that encourage visitors to use natural language search

Such companies are going out of their way to urge visitors to use their sites by asking questions or making requests – 'How do I apply for an internship?' 'Do you test on animals?' 'Show me the latest financial results'. And so on.  

This interrogative approach to information retrieval is basically the same as that encouraged by the new wave of voice assistants.

The problem at the moment is that, like early voice recognition software, corporate sites’ natural language search engines often don't work very well. In fact, most types of corporate site search engine don't work very well (though you can read about some noble exceptions here).

This is one reason why 'traditional' navigation menus are still so vitally important on corporate sites.

But as more people get used to using the internet by talking instead of clicking, they may increasingly expect to be able to do the same on corporate sites.

Navigation buttons will almost certainly not die as a result: even when people stop feeling silly talking to a computer, it will not always be appropriate for them to do so – in a quiet office, for example.

But more people may well want to browse by mouth rather than mouse at least some of the time.

What should web managers be doing now to prepare? Improving their corporate site’s (text-based) search tools is one good starting point. If the search engine is rubbish using text, it’ll still be rubbish using speech.

Ensuring that site section labelling – and use of language in general – is crystal clear is another. If a heading is ambiguous for a human, it's also going to be ambiguous for voice-recognition software (as well as any other kind of search technology for that matter).

And, of course, it will still be important to make sure that your website actually contains the information that people are looking for. Lack of material that provides answers to visitors’ questions is one of the big areas in which many corporate sites’ existing natural language search tools fall down.

Finally, the rise of voice recognition is another reason to ensure that your site meets high accessibility standards: if the site works well for sight-impaired visitors using text-to-speech screen readers, for example, then it is more likely to work well for all users of voice recognition technology. 

Indeed, like other recent technological shifts, the rise of voice recognition doesn’t rip up the existing rules of good corporate web management – it simply makes many of them more important than ever.

- Scott Payton