In the absence of anything shiny, the hoi polloi and the media focused on Siri as the The Big Idea at the iPhone 4S launch. Android has supported basic voice commands for ages, but there are now a number of Siri-wannabes in the Market. But are they of any use other than as a bit of mild amusement? That depends on the use you put …
Is it possible to have an affair with any of these PA's?
More likely with the Android version than Siri - The phones use a standard interface port.
Now what about yours?
The important thing about Siri (and what raises it above the standard voice-to-text converters and voice command systems) is the understanding of context.
I got no impression from these reviews that any of the apps demonstrated that ability. It's one thing to convert a voice sentence to text then pass it to Wolfram Alpha; it's an entirely other thing to maintain the understanding of context across a number of spoken sentences.
One thing that I noticed on my Samsung Galaxy S2 with vLingo is that it does have history context.
If, for instance, you ask it something about baloons and then ask it something about how to make a pancake - it may ask you if you are talking about the same thing!
Please note, I'm tired and it may not have been vLingo but may have been Jeannie!
Siri is not *that* good at context. Try laying in the gutter after being hit by a car and saying "Call me an ambulance".
You're an ambulance!
You got it in one!
Siri costs bandwidth since it runs on webservers/cloud
Siri costs bandwidth since it runs on webservers/cloud... google should rent watson, IBM might not sell its tech. Most people dont realize in addition to voice recognition for commands in android/chrome/voice-search google has its super-set aka "Google Prediction API" ...(ppl are mistaken that its google-voice, goog-411, AI-Car etc)
Clearly not that much context.
The reviewer's example showed siri failing while another android app gave the correct answer.
My hovercraft is full of eels
Been using Iris, mostly nonsensical answers lots of LOLs, but it got one right "how tall is Kevin Keegan".
How about asking the same questions, same requests to do stuff, etc, on all of the apps (including Siri) and then telling us how well each of them performed? Half of each review describes what it looks like, which we can see from the images, which doesn't leave much room for anything else.
It's quite cool to see voice recognition finally finding newsworthiness.
Siri's impressive, but not so much for what it does for me (which is pretty much limited to setting alarms truth be told) but for the fact that it's there without being in the way, so to speak. I guess the same can be said for the alternatives listed here.
The big problem with all of this stuff atm though, is that you still have to press a button to use it.
I think I actually hate myself for even considering posting this ...
Voice recognition = identifying the owner of a voice.
Speech recognition = recognising and interpreting the spoken word.
Siri and friends are speech recognition rather than voice recognition. The distinction appears to have become blurred over the years but it's still quite a useful one.
So very true. Maybe somebody could ask the respective digital assistants the question "are you doing voice or speech recognition", would be a interesting response to have.
On the iPhone you can also just hold the thing up to your ear and Siri comes up, no button needed.
In the long run you'll need a specialized low-power chip that gets programmed with the name you would like to call your phone and that listens all the time, waking up the main CPU as soon as it hears its name. Everything short of that won't work, you can't have your phone listening all the time, this draws much too much power (and there are some creepy privacy problems lurking here too).
We will get there sooner or later, though. For many mundane things speech recognition and AI systems are just the perfect solution.
Iris' answer was "What is your purpose in asking?"
Wolfram Alpha - hah
obviously not that clever.
I asked it a question about converting "metres" to chains and it very promptly gave an answer in "meters" with the result illustrated in terms of multiples of an American football field and a Manhatten city block.
And it can't understand converting a number of megacycles to corresponding wavelength in metres (or anything).
Maybe try an SI unit like Hertz instead?
Any application ... or any person, for that matter ... that can't convert megacycles to wavelength is a complete waste of oxygen! I can't even understand why you would want an app to do something that simple for you ... poor schooling, perhaps? Sheesh! Save the processing power for something difficult, why don't you?
But isn't it San José, California?
Population, a shade less than a million.
Maybe it decided that you meant Costa Rica?
Coincidentally I loaded "speaktoit" on my Droid3 over the weekend. And, wow, it killed phone performance. Everything was slower. Once I removed the app then speed was restored.
I wasn't very impressed with its behaviour, either; giving it one-word answers typically failed to cause a reaction ("yes"), so I had to say it twice which resulted in "yes yes". Dunno if that's an underlying speech recog issue with Android or just "speaktoit" being dumb.
I found speaktoit to be very good, before realising that I couldn't think of many situations where I would prefer to dictate my phone to do something for me... I can't imagine doing it in a crowd of people (probably too noisy anyway).
Gave Vlingo a go...
...as it's native on my SGS2 and I don't seem to be able to get rid of it.
After an hour of speaking clearly, and trying to see what I could get it successfully do (trying to give it things it seemed to want to do like send texts etc.)
I had a pitifully small success rate and I speak with an RP accent. It would get a lot of what I was telling it but stuff it up in such a way that I'd have to go and spend ages correcting it. It can send a simple, and very clearly annunciated text message, but will send to the wrong contact, or it will pull up driving directions but fail to understand where I want to go. If I'm in an area where I don't have wifi or 3g upwards I might as well forget being understood at all.
I like the idea and I'll definitely try again in a year or so when they have these services a little more accurate, but I'm not American so these generally won't work for me. If you leave the little beast on by accident it really sucks down the battery, and accidentally pressing it pauses background audio which is very annoying.
In the mean time if anyone could tell me how to disassociate vlingo from my home key (or calibrate the double tap) without rooting the phone I'd be incredibly grateful. (I save asked Samsung to help with this when they release ICS for SGS2 as well but doubt I'll be heard.)
Plain' ol Google Voice Search does the job for me.
Navigate to Y.
I can't really foresee myself needing the things in the commercials, like "Remind me to pay my credit card bill on the 28th."
Vlingo power use
The reason why Vlingo Safereader hammers your battery is that it heavily loads your network connection; it does most of it's actual audio processing on a central server; see: http://www.vlingo.com/about/technology
So everything you say to it goes via a server in the US or wherever; and it requires pretty much unilateral access to your phone (access all files; all network info; etc..) Ie: a total dependence on network connectivity even for dialling numbers etc. and an ability to gather voice prints plus access almost everything on the phone in a jurisdiction that might not be to your liking. But great voice recognition, pronunciation, and a reassuring privacy statement. Pick your poison.
The Mystery of Edwin
Doesn't appear in the Android shop when using the Galaxy 10. Are they implying it's not ready for that machine?
Speaktoit hasn’t impacted performance on my 2.2 but, boy, does this phone have a problem with my crystal-clear Received English. A shamefully bad impersonation of an American drawl seems to hit the spot, though.
"Call me a taxi."
"Certainly, Dave. You are a taxi."
re: "Call me a taxi."
I'm sorry Dave, I can't do that...
Call me paranoid...
Each of those apps require some combination of these Android permissions -
-Read Phone Status and ID.
-Services that cost you money (make calls, send SMSs).
-Read contact info directly.
Fair enough its conceivable why an app might need access to your contacts, even to make phone calls for you, but the question to ask is whether it makes calls without you even knowing, perhaps at 4am, to their premium rate number in thailand!!??
Quite why such an app would need access to your IMEI or phone numbers is beyond me save for some kind of info gathering and selling scam - something to do with them being free?
And I should apparently check the ratings from thousands of people who have interest only in how the app performs to see whether its safe or not? Yeah right.
For all of Apples many *many* failings, at least you don't have this problem.
More apps to compare
I agree with IR that a consistent test script is very desirable. Also, I'd like to suggest that the application Jeannie (also known as Voice Actions). It has a large list of built-in commands that the others do not seem to have...
- Product round-up Six of the best gaming keyboard and mouse combos
- China building SUPERSONIC SUBMARINE that travels in a BUBBLE
- Boffins attempt to prove the UNIVERSE IS JUST A HOLOGRAM
- Review Raspberry Pi B+: PHWOAR, get a load of those pins
- Linux turns 23 and Linus Torvalds celebrates as only he can