Voice control showdown: Siri vs Google Now, S Voice, BlackBerry and Windows Phone 8
Everyone has heard of Siri, the personal digital assistant hidden away inside your iPhone but there's more than just Siri on the market. With hardware features of smartphones matching each other step-for-step, manufacturers are increasingly looking towards bespoke services and features, such as voice control, as key differentiators.
While third-party developers were already working on voice control apps for handsets it was only since the introduction of Siri (itself a third-party app available on the App Store until Apple bought the company in 2010) that rivals really started to put more emphasis on their own equivalents.
However, voice control is no mean feat, so how does the competition measure up to Siri, the current queen of the crop?
In order to put each system through its basic paces, I tested out an iPad 3, Samsung Galaxy Note II, Nokia Lumia 920 and a Sony Xperia T running Android Jelly Bean.
All voice recognition tests were done indoors in a mostly quiet environment, which makes a difference. Try replicating these on the streets and you'll likely get a worse word recognition rate, and more erroneous actions performed on your behalf. You might also look a bit strange walking down the street shouting at your phone telling it to take you to the nearest Nandos.
Siri, like it or not, popularised voice control and is seen by many as the leading system on smartphones today.
What Siri excels at, particularly in comparison to some of the other systems in this test, is understanding natural language.
For example, Say "I want to take a picture" and Siri will open the camera app or you can just say "open camera".
It's also really good at recognising what you're saying and giving you pointers along the way if you're new to the phone. For example, you can send email using Siri but if you haven't set up your account first it will tell you to do that, rather than just return an error.
The range of functionality with Siri is quite broad but dependent upon the key partnerships Apple has put in place, so can vary depending on what you are trying to do. It's tied into most of the core apps in iOS so making or changing a calendar entry, sending a text, or getting information about a particular event is a breeze.
However, at points the information is very superficial, for example, asking when the next Formula One race would take place returned the correct answer immediately but when I asked where it was, Siri just repeated the date. Similarly, when I asked when the last race of 2013 would be, Siri said it could not get information about Formula One and offered to search the internet instead.
One of the strengths of Siri is the tie-in with Wolfram Alpha giving great results for computational answers. Ask Siri a maths question and the answer will pop straight up but ask it to switch on Bluetooth and it can't help you out.
When I asked Siri to "take me to London Bridge" it replied by popping up information about London Bridge, as well as saying how far away it was.
Siri's biggest weakness (in the UK at least) for me is the local information. Ask where the nearest Nandos is and it says "I don't know what that is" and offers to search the web.
You can also use Siri for things like taking notes, playing music or videos, checking the weather or listening to notifications.
While Android has included voice commands for some time, the most recent version – Jelly Bean – also has Google Now, which isn't quite the same thing as Siri or S-Voice as it's based around Search and other Google services. Nor is it the same as the other stock Android Voice commands system that focuses more on hardware control. That's not to say Now can't be used for handset control functions too, although it is a little more limited in this department than some of its competitors.
For example, while you can say "where is the nearest Nandos?" (it gave the best answer of the bunch, pictured above) or "navigate to nearest petrol station" and it will do what you ask, you can't carry out system functions such as turning Wi-Fi on or off. I could only seem able to open certain Google-made apps through Now. Given the integration with things like Google Maps, telling it to take me to London Bridge automatically popped open Google Navigation with a route all loaded.
The level of voice recognition and understanding of Google Now was very impressive, with it easily understanding individual words (particularly ones that are easy to misinterpret) or slightly vague questions, such as will I need a coat tomorrow, which it answered with ease. Google Now's voice recognition was easily the best overall at understanding what is being said.
Google Now, is in some ways limited by its lack of ability to perform some system functions, but it can do things like open third-party apps that you've installed on the phone.
Open Now and ask it to send an email, call or text one of your contacts and it will have no problem understanding the words, but bizarrely – in my testing at least – word recognition when asking it to add one number to another was atrocious; it simply couldn't understand the word 'add' or 'sum'.
If I told it divide, multiply or work out the square root it got there first time, though.
Strangely, while newer Android phones have Now and the older Voice Control both apps are accessed separately.
S Voice is Samsung's voice control system found on handsets, like the Galaxy S3 or in this case its Galaxy Note II 'phablet'.
Like the other systems in the test here, S Voice is well skilled in performing tasks like switching Bluetooth on or off on the phone, or sending a text message to a contact. Weirdly though, the S Voice system can't seem to send an email. You can also use it to update third party apps like Twitter or Facebook, though.
On the occasions that S Voice doesn't quite hear the instructions clearly (or correctly) it will search for the nearest match to what it heard. The system also does better than some of the others in the test with things like scheduling. For example, if you try and schedule an appointment for the same time as another that is already in your calendar, it will warn you and pop the other meeting info to check you want to schedule two things for the same time. The BlackBerry voice system doesn't do this, but Siri does.
S Voice was also one of the only other systems (in addition to Siri) to correctly put a reminder in the calendar when asked to "buy flowers on Valentine's day". It also did pretty well on the Nandos test – asking where the closest was automatically resulted in a web search, though there was no mapping result to be found.
However, asking S Voice "how to get to London Bridge" resulted in Google Navigation popping open an address box that required pressing a button to select the precise destination.
For me the voice recognition seems to be a bit hit-and-miss with S Voice: on some occasions I was impressed that it could pick up what I wanted it to do, but in others, it repeatedly got things wrong, such as when using the calculator. It also doesn't seem to understand computational function, as when it did finally recognise me telling it to add 460 to 320 it responed: "I don't know if I can answer that properly, do you want me to search the internet for 460 + 320?".
BlackBerry's voice control system on the BlackBerry 10 OS is the new kid on this block.
Like Siri it can be used for things like sending text messages, placing calls or scheduling a meeting or reminder. It can also be used to take notes or to dictate other text.
Word recognition on the BlackBerry system is actually pretty accurate — most of the time it seems to understand what is being said if you speak clearly — but it can't understand natural language in quite the same way as Siri and Now. For example, if you say 'take a picture', the handset responds with 'do you want to search the internet for take a picture?' However, say 'open camera' and it can do that.
Like Siri, the BlackBerry system asks for confirmation of commands. I asked it to send an email to Ben Woods and it responded by asking which Ben Woods email address in the address book I would like to send it to, all achievable through voice commands.
It also has quite a granular level of control of the information being entered or edited, for example, when scheduling an appointment or writing an email it will allow you to select and edit each part of the message (time, title – for an appointment, or things like subject line, body, recipients for an email) without needing to touch the handset.
It did occasionally find it difficult to recognise confusable commands. Such as, instructing it to set a reminder that it was Ben Woods' birthday tomorrow ended up in a calendar entry saying 'Ben would birthday'. It's only a minor point, but worth keeping an eye on if you're sending messages to people that have similar sounding names.
Asking the BlackBerry Z10 where the nearest Nandos is resulted in the phone offering to search the web for the closest branch.
Where the BlackBerry falls hardest is the lack of a partnership with a service like Wolfram Alpha, as offered by Siri. Asking it any kind of mathematical addition or division simply resulted in it offering to search the web, despite understanding every word, rather than displaying the result.
However, the voice commands are pretty powerful overall. You can open any app on the phone or search for any phrase, contact, word or anything else all via search and it will understand what you are saying quite a lot of the time. And if social networking is your thing, you can update your status on Facebook or post a Tweet without using your hands.
Windows Phone, like all the others, offers a voice control system but without some of the finesse or detailed features of other systems.
For example, asking it to text a contact in the contacts list returned the correct result without a problem, even prompting for the dictation of the rest of the message and confirmation of whether I wanted to send it. Google Now, by contrast, simply pops up the composition window. However, actually trying to use the system to dictate a message was incredibly frustrating.
Overall, the Windows Phone voice control feels a little less 'intelligent' than some of the others – asking it questions that you might ask any of the other systems usually results in a web search. For example, ask it to turn the Wi-Fi on or off and it can't help you, but it's bizarrely accurate if you use it to open apps on the device.
Think of the Windows Phone voice control system as less of a virtual assistant and more of a voice search tool – nearly everything you do will result in a Bing search, anyway.
Update: Some comments below pointed out the power of Windows Phone's app control system, so I went back for another look. Sure enough, while Windows Phone 8 did pose some frustrating problems in areas where others excelled it does offer features that the others do not.
For example, saying "open BBC News most read" resulted in the BBC News app opening and then the handset read aloud the headlines of each of the ten most read articles. Other apps also support this kind of action, with Audible allowing you to go directly to your library or specific audiobook by saying "open my Audible library", for example.
However, this functionality seems to be app specific and similar tests on apps such as Epicurious and Amazon could do no more than simply open the app.
Voice control has been around for a good while now, so it's frustrating that it still isn't perfect as that's what it really needs to be if it wants to be a viable way to interact with your phone in a dependable way. You wouldn't be very happy if half the time you pressed a button it did what you wanted and the other half it did nothing, or worse, something you actually didn't want — and sadly that's a big part of the experience with voice control.
Using the software indoors (where these were tested) improves the chances of it doing what you want, and out of the bunch Google Now was the most useful and best at recognising what was being said, that's not to say that Siri has nothing to offer — clearly its partnership with Wolfram Alpha is one of its strongest points, and a factor that other systems would do well to provide in some way.
As for Windows Phone's voice control in my experience it's just not good enough to be useful or reliable — although it is a handy way to quickly open apps.
BlackBerry's system sits somewhere in the middle of the pack, it has hardware functionality like being able to switch Bluetooth on or off, but its actual recognition is a little hit-and-miss at times: still, it fared better than Windows Phone 8.
The question isn't really which voice control system is better, right now it's more about which has the biggest limitations.