Google 'accelerating' development of its own voice assistant

[monkeyluis]

http://www.androidcentral.com/google-accelerating-development-its-own-voice-assistant-wsj-reports

Hmmmmm, sounds familiar.

Let's see what comes out of it.

 

[Shadowriver]

To make it visibly better then Siri they need to let developers contribute to possibilities of actions to that assistant

 

[Maikai.Guy]

To make it visibly better then Siri, all it has to do is work. ;-)

Voice recognition/transcription is an incredibly important feature for any digital assistant. Siri is a great attempt by Apple to not only begin to offer it, but to leapfrog the existing offering from Android, and I applaud the direction. I look forward to the competition, because when companies try to leapfrog each other, the users of both platforms benefit.

 

[GMJeff]

But, seriously, how many actually use Siri or any form of voice assisted function on a smartphone.

I personally find it easier to type a text or a search query or a phone number than telling the phone and getting it all messed up, then trying again. Sometimes numerous times.

Apple is pushing Siri to a crowd of people that may have problems understanding a UI on a phone and tries to simplify the device in such a way that "all you have to do is talk to it" and it does all of your bidding. It is a gimmick that will sell millions of more phones.

I don't buy my phones based on how well it decyphers my voice inputs, but rather how well it makes phone calls and runs regular apps that I use on a daily basis.

I don't see voice apps as being a pertinent and reliable enough item until at least a year from now.

 

[Maikai.Guy]

I have used the voice commands and the voice to text capabilities of Android for years. The streaming voice to text of ICS was a welcome improvement.

What you suggest is kind of like saying people who use cruise control in their car must have some problem using their gas pedal. Voice commands and Voice to Text, like Cruise Control, are an added feature and benefit that makes operation easier. Neither is necessary.

And where does this argument stop? If we use our remote controls for our TVs, does that mean we must have trouble using the controls on the TV itself? ;-)

It sounds like you simply haven't integrated voice commands into your daily usage and then assume everyone is like you. I'm guilty of thinking everyone is like me. I do it all the time.

 

[Shadowriver]

To make it visibly better then Siri, all it has to do is work. ;-)

Voice recognition/transcription is an incredibly important feature for any digital assistant. Siri is a great attempt by Apple to not only begin to offer it, but to leapfrog the existing offering from Android, and I applaud the direction. I look forward to the competition, because when companies try to leapfrog each other, the users of both platforms benefit.

Existing offering of Android is just Voice Recognition which could be used to voice commands and it's nothing new, but it's very low level implementation, it only converts voice to text, how you analyze it your job as developer. Im not sure in Android had assistant apps before Siri, but is next level and it implements converstive way of interaction in one place. What Apple dont have is API which would able to extend possibilities of Siri infinitly, if Google will do that Siri is done and Apple will need to do something more as responce. But i kind hard for me to imagine how such high level feature can have API... but google will figure something ;]

But, seriously, how many actually use Siri or any form of voice assisted function on a smartphone.

I personally find it easier to type a text or a search query or a phone number than telling the phone and getting it all messed up, then trying again. Sometimes numerous times.

Apple is pushing Siri to a crowd of people that may have problems understanding a UI on a phone and tries to simplify the device in such a way that "all you have to do is talk to it" and it does all of your bidding. It is a gimmick that will sell millions of more phones.

I don't buy my phones based on how well it decyphers my voice inputs, but rather how well it makes phone calls and runs regular apps that I use on a daily basis.

I don't see voice apps as being a pertinent and reliable enough item until at least a year from now.

Open your mind There 3 core audience for voice control in any form

-Car drivers, which have busy hands and should have busy eyes
-Headphone users to which having few functions on headphone buttons is not enoth
-People which have problem using touch screen, have hand problems or bad/no sight

And heres demo of that here
For lot of people it might look like a gimmick.... and thats absolutely normal because you they not target audience of this feature, no body force you to use it and it not hurts you that this feature exists on your phone, same as many other features that you don't use. I bet if Google would announce Siri like thing first you would use it as argument against iOS.

 

[monkeyluis]

I have used the voice commands and the voice to text capabilities of Android for years. The streaming voice to text of ICS was a welcome improvement.

What you suggest is kind of like saying people who use cruise control in their car must have some problem using their gas pedal. Voice commands and Voice to Text, like Cruise Control, are an added feature and benefit that makes operation easier. Neither is necessary.

And where does this argument stop? If we use our remote controls for our TVs, does that mean we must have trouble using the controls on the TV itself? ;-)

It sounds like you simply haven't integrated voice commands into your daily usage and then assume everyone is like you. I'm guilty of thinking everyone is like me. I do it all the time.

Accessibility is necessary.

 

[monkeyluis]

But, seriously, how many actually use Siri or any form of voice assisted function on a smartphone.

I personally find it easier to type a text or a search query or a phone number than telling the phone and getting it all messed up, then trying again. Sometimes numerous times.

Apple is pushing Siri to a crowd of people that may have problems understanding a UI on a phone and tries to simplify the device in such a way that "all you have to do is talk to it" and it does all of your bidding. It is a gimmick that will sell millions of more phones.

I don't buy my phones based on how well it decyphers my voice inputs, but rather how well it makes phone calls and runs regular apps that I use on a daily basis.

I don't see voice apps as being a pertinent and reliable enough item until at least a year from now.

Think about it from someone's perspective who has a disability.

 

[Mercury81]

I can hear it now "Hello today is Tuesday, the 25 of June year
2012, thre current weather is sunny with a temperature of 83 "

from Evo4GLTE on Android Central

 

[GMJeff]

I never said that voice command doesn't have mainstream applications now that can be used for advanced control of devices, I was merely saying that people that use the voice command features are a small percentage of overall users. People with disabilities as well as voice commands for motorist hands free devices, as well as some other functions.

Apple is trying to push a system on their phone that does not sell a large number of cell phones. They seem to be trying to convince users that Siri will somehow make their life easier by sitting in a chair and asking the phone to tell you a joke.

As has been said many times on these forums is that Apple buys or takes something and tries to make the world think that they need to have this. Voice recognition and commands have been in devices for years, as far back as some of the early Windows Mobile devices (PDAs), but have yet to catch on mainstream in a way that people can't live without them.

As far as the poster that told me to open my mind, I have owned all types of devices, from the aforementioned Windows Mobile PDAs, Blackberry, iPhones, Androids and Windows Phone devices. Frankly, none of the devices that have had voice control/commands have been of any use to me, as I can use the device quickly without the need of asking it to do it for me.

But as I said before, I just don't believe that the voice command systems will be more mainstream until about a year from now. With quad core chipsets and LTE cell service coming into its prime, the pieces are all falling into place to allow users to ask the phone a question and have it return a query response in a timely manner for internet related searches. Local queries on the device should become seamless and quicker as well.

 

[anjali_jain]

Siri now works, and works well? Have you talked to iphone consumers? Every person I know who contain it hates it.

 

[GMJeff]

Well, there is the problem, you are asking the consumer, not Apple. That is a no-no. Apple wants to tell you that it is better than sliced bread and you can't live without it.

I have used Apple products (MacBook Pro, iPhone 3G, 3Gs, 4 and iPod Touch 1st gen) and frankly can see nothing special about them. They do the same job as a PC or other smartphone that costs less.

There were some things I liked about the MacBook and the iPhones, but the shine wears off quickly when you can't get a piece of software that you used to use on your PC all of the time for the Mac. Then you use BootCamp to run PC apps. Then you realize you are booting into BootCamp on a regular basis and only booting into MacOS when you need to do remedial tasks. The $1500 spent on the MacBook then seems like a burden. I can buy 2 PC's that have better hardware specs than the Macs and still have money left over to buy Office or some other software for productivity.

I know I sound like I hate Mac and I don't, the logic just escapes me as to how Apple makes commercials and people run to them in a blind need.

I still remember a commercial that Apple aired in 1984, if I remember right. It was a take on Orwell's idea of 1984 where a lady is running with a sledgehammer and eventually throws it through a jumbo screen depicting "Big Brother", a video image of a man's head that told people what to think and how to act.

Apple tried to make itself into the rebellious image of a company that was fighting against the large corporate machine (at that time it was IBM, if I remember, Microsoft didn't come until later). They were there to break everybody out of that corporate mold and allow everyone to "Think Different".

The problem that I see is that Apple seems to have turned into the "Big Brother" image, as everyone seems to want to copy Apple's designs to try to steal customers away.

I know that this doesn't have a lot to do with this topic, just needed to rant a little, sorry.

 

[DroidDavi]

I really hope google pulls this off. They have the data to back it up i want it to work. I more and more use the voice features on my Gnex and my Kinect and would love to see something like this done right. I bought and use Nexus phones for this waiting on that new feature come on JellyBean

 

[Shadowriver]

By opening your mind i didn't ment that you not used other devices, but you sound like you don't understand role of voice assistant, lot of people seem have problem on understanding role of new features, even Android users. Yestoday i seen comment of a guy that see pointless to place NFC on Wifi only Nexus Tablet, so i showed him this:

Microsoft's Windows Phone Summit & Windows Phone 8 - YouTube

+ it can be used to quickly negotiate data transfer... it can be used for anything, NFC is just a medium, the key here is to build userbase no matter if its just wifi model, or else NFC is useless. Lot of people critisize new feature in closed mind and don't think about possiblities, if everybody think that way we would stay in stone age

But back to topic

You said something like this:

Apple is pushing Siri to a crowd of people that may have problems understanding a UI on a phone and tries to simplify the device in such a way that "all you have to do is talk to it" and it does all of your bidding. It is a gimmick that will sell millions of more phones.

Is defently not a gimmick for reasons i said, it add value to the product and as i showed you on video thats not a a main target. Other 4S ad (since i think you tlaking about them) shows mainly a Siri since thats main new feature of 4S, how else you would advertise it? thats not only new thing in 4S but it's hard to adrvitise just by saying it has better hardware or better camera. But they not making it a main way of interaction as you said, iOS is still iOS and it's system mainly for touch screen.

And yes voice recognition will be better overtime with better hardware.... so why to stop development now? we will have recognition ready for better hardware WP7 made step forward then Siri made step forward so maybe Google will do next step, thats now technology drives as Maikai.Guy said.

 

[Maikai.Guy]

Existing offering of Android is just Voice Recognition which could be used to voice commands and it's nothing new, but it's very low level implementation, it only converts voice to text, how you analyze it your job as developer. Im not sure in Android had assistant apps before Siri, but is next level and it implements converstive way of interaction in one place. What Apple dont have is API which would able to extend possibilities of Siri infinitly, if Google will do that Siri is done and Apple will need to do something more as responce. But i kind hard for me to imagine how such high level feature can have API... but google will figure something ;]

I don't agree with much of what you said.

"Nothing new" I kind of agree with, only because Android has had it for years, so it's nothing new for Android. As far as I'm aware, although people had been playing around with voice recognition for years, it was never more than a niche application. Android was really the first broadly distributed example of using Voice commands and voice recognition for general society.

In Android, we have always been able to ask "Definition of street car." and have Google go search the web for definitions. Or ask "Nearest Italian restaurants." and get a list. Or say "Navigate to XYZ electronics." and get turn by turn directions. I can continue, but won't.

As for the API, it's already there. APIs are calls to functions. When I say "Navigate to xyz electronics", the term "Navigate" is recognized by the voice translator as a function call to the navigator function. The voice translator continues it's translation and passes on the remaining data as the variables for that call.

I see SIRI as no more than a minor improvement in the human interface aspect only. Questions can be asked in a more natural way. That's all. Other than that, Siri actually offers less functionality than Android's "very low level implementation", as you claim. I say this because Siri craps the bed... a lot. My nieces are iPhone fangirls and they've shown me Siri from top to bottom. 1 time out of 5, Siri doesn't know what the heck they asked. And when trying to use Siri to dictate, it is rife with errors.

But again, I like the direction and I think competition between Apple and Google regarding the voice interface is an exceptionally good thing for people like me, who use it frequently.

 

[Shadowriver]

Thats what im saying Android was really the first broadly distributed example of... nothing new. By low level of implementation i meant it just got voice recognition, thru i must admit i didn't notice voice comands in search bar until yestoday when i tied to switch language , but i bet Google search bar is not first voice command app like that even before android. Inclusion of understanding natural talk it's not a minor improvement imo, it's not easy to implement something like that.

API is not there, what you see there is just few commands that makes specific intent calls, but it can't go beyond that, so that not API or rather it's google search bar that use APIs of other apps not other way, which is no diffrent from what Siri actually doing (ofcorse it can't use 3rd party apps with same functions as like in Android). Twitter didn't add "Tweet Something" because they lazy... they simply can't, they could plug in to Search intent but that would mess up whole search in system. I was thinking about API which would allow add commands free something like url opening intent but for voice commands (or rather text commands). It's easy to implement for prefix commands as Android uses (Android already can find which app to use from url fragment) , but harder on natural talk it's not so easy since it's hard to identify command and send it out to right app or else we gonna go back to simple one.

 

[GMJeff]

I don't mean to keep reiterating my point, but I see the uses for voice control/command systems today in niche markets.

I also say bravo to Google for trying to advance this ecosystem also, as they are trying to better an application which will allow them to help sell more tablets and phones as well as chromebooks.

All i was getting at is that when the public finally sees that this is what they really want, Not what Apple is trying to shove down their throat, Google will be ready for it, but it will be at least a year from now before it can be implemented properly where there are minimal amounts of errors on the software voice recognition side.

Dragon, who is supposed to have one of the premier voice recognition suites for a computer still gets things wrong on a regular basis. Some of that may be the way people talk as well as accents which can make one word sound like another.

My whole point about using other devices was that I have used many phones and computers and never had to use a voice command system. I have tried them, but find them somewhat archaic. The recognition algorithms are just not totally refined yet. Manual entry in terms of a styles or touch are quick if you know how. Swipe (swype?) keyboard would not be as popular as it is if people used voice recognition software all of the time to enter data.

When they are better, i will try them again and see if they make my life easier. Until then, someone needs to spend more money on development. It appears that Google is that someone.

Sent from my SAMSUNG-SGH-I717 using Android Central Forums

 

[Maikai.Guy]

Inclusion of understanding natural talk it's not a minor improvement imo, it's not easy to implement something like that.

Google understands natural talk. They have for years. It is not a simply voice to text application. It understands context. If you say "I'll see you next week", it will use the word "week." If you say "I haven't eaten and I feel weak", it will use the word "weak." Try it for yourself. Use other examples. It can only do this if this if the translator understood context.

API is not there, what you see there is just few commands that makes specific intent calls, but it can't go beyond that, so that not API or rather it's google search bar that use APIs of other apps not other way, ...

I'm not sure what you're getting at. API support is clearly there. An API is a function call. It is never contained in the function being called. It is an element of the routine currently running... in this case, voice to text and the Google Search bar. It calls the target function and passes variables to it. Why do you believe that any of this needs to resident in the called function? I take it you don't do any coding.

As for people implying voice commands and voice-to-text is mostly for handicapped users or a minority of users, I disagree. I don't believe a majority of users take advantage of it, but it's not a small minority by far. I use it, as well as many people I know. Voice commands and voice dictation is far faster and easier than typing in the same thing. I can call my wife by touching one button and saying "Call [blah-blah]." I can navigate to a company or person I'm visiting by touching one button and saying "Navigate to xyz." I can reply to e-mails or texts by touching two buttons and speaking my answer. ICS made a huge improvement in this by supporting streaming voice to text. Context if fully recognized and a mistake is a rare thing. In fact, I no longer proof read what I dictate. In the rare case of a mistake, it is almost always very minor.

 

[Shadowriver]

Google understands natural talk. They have for years. It is not a simply voice to text application. It understands context. If you say "I'll see you next week", it will use the word "week." If you say "I haven't eaten and I feel weak", it will use the word "weak." Try it for yourself. Use other examples. It can only do this if this if the translator understood context.

By natural talk i mean more human way to ask not saying static commands.... but thats does not matter now since it can now handle that

I'm not sure what you're getting at. API support is clearly there. An API is a function call. It is never contained in the function being called. It is an element of the routine currently running... in this case, voice to text and the Google Search bar. It calls the target function and passes variables to it. Why do you believe that any of this needs to resident in the called function? I take it you don't do any coding.

API is application programming interface, as name states it's a interface such as function, not a call of that function

Ok maybe i didnt said it clear, i mean Search don't have API which let you extend it functionality. Yeah android got speech to text that let you fo voice commands but developer of app is the one who need analyze text in this situation. Voice command in this from don't encourage developer to implement that.

I assume we talking here about one place to issue commands without using other 3rd party software sane as Siri do, in this case Google search bar. Currently what is do (or did in 4.0 and below) is react to specific hard coded commands and triggers so called intent (Intent | Android Developers) assigned to that command which searches for appropriate application that can handle this intent, for example when you say "navigate to" it will trigger navigation intent with argument containing destination that you said and open right app or ask if there more app that can handle that intent. But this uses API of other applications, which been used even before voice commands in Android, search text does not have API. You can't extend functionality of text search or you replace search bar all together with more commands. It's no diffrent from that Siri is doing now too (thru i bellive some things are hard coded in to Siri it self)

My idea is that Search bar API oreven other apps like that is to find app that can handle text commands (so they can be not only used in voice recognition) like it's doing now with URLs with open file intent for example. So for example Twitter have defind in manifest that it can handle text command intent with filter on prefix "tweet" and it will open Twitter or get special widget for search bar or anything else that let you send tweet that you said. And thats only my example, i know Google could do something a lot better. Simple commands are easy to handle like that but in natural speech where command don't need to be on begining it's hard to filter down what user really want from manifest.

Yea Intent system makes what i said kind of look no different from what it's doing now in technical stant point, but thats how app to app comunication in android works. If android would use app to app communication using normal function calls, app could for example register voice commands to Search Bar or voice recognition all toghther, but android intents are more passive way to doing so by defining it in application manifest which is in fact better and more open.

Now what happened today, they announced ability to use Knowlage Graph which makes search bar make it feel like siri, but still you can't expand functionality of it with our own app. In fact there was a question on Android Firechat session if the new stuff in search bar got any APIs and they said simple "No". But still it's kind of better then Siri, Siri can only view Wolfram Alpha result when new search bar can directly answer question on scientific stuff

 

[Maikai.Guy]

Sorry Shadowriver. I'm having trouble following your writing.

I think I disagree with at least half of what you wrote. ;-) ...but it's hard to pull out your points and address them.

Clearly I misunderstood what you meant by "natural talk." I thought you meant context recognition, which is arguably most difficult. Understanding context of the speech (proper translation of there or their... week or weak...), on the other hand, requires a bit more processing.

Siri doesn't actually "understand" natural talk. The user physically triggers Siri and then asks a question. An algorithm simply selects the operative words from the sentence, calls the appropriate function and passes the variables . Once a search result is obtained, the Siri program parrots/rephrases the operative words and supplies the "headlines" of the results in voice. This is very similar to what Android has done for years... without the voiced response.

Go ahead and add "filler" words to a voice request, like "Please." If you natural talk ask "How do I get to xyz?", search will provide web results for getting to xyz. If you pull up Navigator and natural talk ask "How do I get to xyz?", navigator will route you to xyz. You'll find Android dismisses these "filler" words and essentially understands "natural talk."

What they did with Siri is expand the algorithm and added the voiced search results. Definitely a nice improvement, but a fairly minor improvement, at best.

The improvement is lost though, because Siri currently lacks reliability. I'm sure they will improve it, but it's a work in progress, whereas voice commands and voice to text is very solid on the Android platform.