Siri? Isn’t that Tom Cruise and Katie Holmes’ firstborn?
No, no, you’re thinking of Suri. I’m talking about Siri. Or rather, I’m talking to Siri. And Siri is talking to me.
Siri sounds talkative.
Indeed, that’s because Siri is a voice assistant application that comes with Apple’s latest iPhone – the iPhone 4S.
What’s a voice assistant?
It’s a piece of software that responds to verbal commands. Siri is a natural language processing (NLP) app that can recognise and respond to speech. It responds both verbally, via synthesised speech, and where appropriate by pulling up and displaying relevant data on the user’s iPhone.
Examples of actions Siri can perform include taking dictation, sending emails and text messages, scheduling appointments and reminders, locating information, playing media and making phone calls. Apple says Siri uses almost all the built-in apps on the iPhone 4S – plus data from the web – to “find the information you need”.
Siri’s technology is reportedly powered by US company Nuance, another speech-based interface tech company, under a licensing agreement with Apple. It’s worth noting that Siri is a beta app at present and currently supports only three languages: English, French and German.
But what powers Siri?
As well as NLP tech, servers on the back end and the phone’s hardware itself, Siri makes use of Apple apps such as its weather app and a local search data service from Yelp in the US to summon the answers its users seek. Siri also reportedly utilises the Wolfram Alpha search engine to help it respond intelligently to factual queries.
Wolfram Alpha utilises organised data and a variety of algorithms and equations to generate relevant answers to text-based queries. Ask it what the GDP of France is, for instance, and it will return that figure, along with a variety of associated data such as the recent history of France’s GDP perhaps plotted as a graph. Siri similarly aims to be so smart – but more mouthy.
I don’t have an iPhone 4S – can I get Siri on my iPhone 4 or 3GS?
Not from Apple itself, although various people have been trying to hack Siri onto other iOS devices. But via official channels, the app is iPhone 4S-specific for now.
Can I tell Siri to fire up my psychedelic disco flashlight app without me having to get out of the lotus position then?
Not at present. Siri only works with native iPhone apps such as the iPhone alarm clock app or calendar. It doesn’t work with third-party apps because Apple hasn’t opened Siri up to its developer massive. If it did that, the possibilities for talking to your phone really would be – ahem – massive.
How long has Apple been cooking Siri up then?
Well, we know that last April Apple acquired Siri Inc, a company that had launched an eponymous voice-activated personal assistant iPhone app – which Apple subsequently removed from the App Store as it worked to bake Siri into the 4S. So the iPhone 4S’s vocal sidekick has been in the pipeline since at least 2010.
Apple has had a more basic voice control feature in iOS devices since iOS 3.0, however. iPhone 3G S, iPhone 4 and iPod Touch owners can access this feature by holding down the home button to activate the system and then telling their device to ‘call Jony Mobile’ or ‘play artist The Beatles’. The system can also respond to being asked ‘what time is it?’.
This voice control feature is much more…
…basic than Siri – requiring voice commands to be spoken precisely to be recognised by the system, and offering only limited functionality to connect calls or play music. Siri is much more capable – hence the need for additional service plug-ins.
In the US, Apple has agreements with third-party local maps and search providers to support a local look-up service. For instance, tell Siri ‘I need a haircut’ and it could pull up a list of barbers in your immediate vicinity. However, Apple does not yet have a deal with a local data provider in the UK – so no local look-ups for Siri users on this side of the pond.
On the capabilities front, Apple says Siri’s got artificial intelligence smarts – claiming it will get better over time as it becomes accustomed to your voice and the kind of things you’re looking for. Apple also says Siri is “proactive” – meaning it will question its owner until it knows what you’re looking for and can act accordingly, rather than giving up at the first incomprehensible word.
The result, reckons Apple, is a more naturalistic dialogue. “Ask Siri to do things just by talking the way you talk”, the Apple website says, adding: “Talk to Siri as you would to a person.” This NLP-plus-AI approach makes Siri different from Apple’s earlier mechanical voice controls – which basically just involved a human repeating pre-set commands at a microphone.
But aren’t there other voice assistant apps out there? Doesn’t Google’s Android already do something similar?
There are indeed other NLP apps. There are other voice assistant apps too such as Vlingo, which offers apps with similar functionality to Siri for various smartphone platforms including iOS. Vlingo also lets users connect to some third-party apps, as well as native phone apps, so users can speak a tweet or tell their phone what to update their Facebook status to – something not currently supported by Siri. Vlingo also allows users to open apps by voice command – another Siri no-no.
Google has also been investing in voice tech for a while and its Search app allows users to speak search queries instead of typing them. Google Translate also lets users speak words and phrases they want translated. Meanwhile, Samsung’s Galaxy S II Android-powered smartphone comes loaded with a Siri-esque virtual assistant – in the form of the Voice Talk app, powered by Vlingo.
Apple is by no means the first to the punch when it comes to NLP, but it is making a big push on the voice front by integrating Siri into its latest iPhone. Expect voice smarts to be a key component of all future iPhones.
Alright, so Siri looks pretty cool but speech recognition tech has been around forever – why the big fuss? Is it just because Apple’s got involved?
Certainly, Apple getting involved does signify a certain level of maturity has been reached. NLP tech is a lot better…
…than it used to be – to the point where it stops being frustrating and can start becoming entertaining. Smartphones with increasing amounts of processing power and server clout on the back end are also helping to smooth the rough edges of NLP by speedily crunching the necessary data.
Because of increased capabilities and usability, voice-based interfaces have the potential to become disruptive, next-gen user interfaces in their own right – offering a new way to interface with devices and search for data. Sure, speaking to your gadget is not going to be appropriate in every situation – it’ll get you kicked out of the local library – but in many instances it’s much faster than typing on fiddly keyboards.
Could a Siri-style interface replace the search engine as the first port of call for web users seeking data? Users don’t care about the process of searching, they’re just after the data, so as Rich Mogull argues in this thought-provoking article, if the path of least resistance to that data is via a virtual assistant, there’s no reason for users to visit Google or Bing half as often as they do now.
Siri-style interfaces could see search engines cut out of the loop for a large chunk of queries – changing the modus operandi of much web navigation.
Interesting. Anything else I should think about?
Well, there’s privacy. If you own an iPhone 4S just stop and think about how much Siri knows about you – stuff like where you live, where you work, who your significant other is, the name of your boss… “From the details in your contacts, it knows your friends, family, boss and co-workers,” says Apple’s Siri website.
But not everyone views these smarts as an unequivocal boon. Illah Nourbakhsh, professor of robotics at the Robotics Institute at Carnegie Mellon University, warns there is a risk of Siri knowing too much – gaining too much intimate knowledge – about its user.
“The [Apple] marketing department doesn’t even get that there’s an issue with privacy that might come up here as you create a proactive agent who knows everything about you, has access to all your apps and knows who your boss is,” he warned. “The problem we have is there is no design ethos vis-