He's completely right though. We've had accurate voice recognition on phones for years.
Siri is only doing a few things differently. This includes:
-Contextual commands with a little more variety than the competition. Eg. "wake me in one hour"
-Some funny responses when you ask it weird questions
-Deep integration with system applications.
Other than that, Siri is nothing new. This should by no means be a selling point for anyone to buy a phone.
Ask it "How to tie a tie"
It returns with a google search. Just with a different interface. It will not actually tell you.
Aside from the commands that interact directly with the OS, that's all it really is. It's a Google search that presents the information neatly.
The voice technology is not new, or any more accurate than other current voice detection software. All the phone is doing is converting your voice to text. The text is sent off to Apple's servers to be processed, and for a command to be sent back. There are trigger words that define the context. Eg. "where is", "wake me up", "what's the weather", "play this album".
Is it currently the best voice command software on a phone? Yes.
Is this amazing new technology? No. But props to apple for integrating it deeply and making it easy and intuitive. But in the end, Siri is either controlling a limited amount of system apps, or it's doing a Google search and displaying the most relevant info based on your search
The difference is one of scale and opportunity.
When Siri is actively running on hundreds of millions of iDevices, and being monitored for accuracy, performance, unfulfillable requests (in multiple languages) -- it will likely generate a greater mine of information than all the prior efforts combined.
This will provide the opportunity to determine what is being done well, done badly or not done at all -- and too flesh out Siri to be even better.
If Apple exposes Siri's capabilities to 3rd-party apps and developers... we will have hundreds of thousands of talented, knowledgeable technical people testing and contributing to this rather specialized subset of AI.
I think that Siri is bigger than a single class of device, a predefined set of tasks -- or a single company.
you: "Siri: text my wife I love you and I miss you"
you: "Siri: send it"
not
you: "Siri: open the messages app"
you: "Siri: select new message"
you: "Siri: find lucy"
you: "Siri: select lucy"
you: "Siri: skip the subject"
you: "I love you and I miss you"
you: "Siri: send it"
Ah, a much better example. I can see some fringe cases for how simply opening the app would be all that is needed, but I have a hard time imagining Apple releasing an API for a small group or users with a high chance the majority could abuse it.
It's funny how now we get people saying how Siri is nothing special, nothing new, nothing that wasn't already available and completely obvious.
If I understand Siri correctly, a single Siri input starts with speech-to text (licensed from Nuance), then Siri's NLP (natural language processing) of the intended meaning, then converting that intended meaning to commands that are then made into specific queries outside of Siri that are then recreated in the most human readable form possible, or executed locally then displayed for you to confirm, and finally text-to-speech. All of these technologies have been around for decades, especially text-to-speech, but I can't think of one that ever been this accurate, useful or seamless. But it's all just obvious...
I haven't gotten through the comments, so it may have been stated already, although I'm not planning to wade through the fan boy hate.
It would seem to me this is a possibility. If Apple only wants this on 4S phones, my understanding was Siri goes out to the Apple DC. If that communication carries with it the model, it would seem easy enough to deny traffic from non-4S models and if it carried with ie the IMEI or other identifying info, it could disable the service or other Apple provided services. The latter may introduce privacy concerns and a legal battle with consumers it doesn't want to take up, but the first seems easily doable.
I'm not for it either way, but I believe it would be entirely possible and within their rights as the DC host to do so.
I think this is the first time Apple has put beta software front and center in an ad campaign. Maybe they are just trying to deflect any criticism of the lack of functionality by calling it beta. In fact I cannot recall any other instance when Apple released beta software publicly. This is clearly version 1.0.
When Google makes a general public software release and calls it beta, Apple fans joke how lame that is. But now, it seems they are ok with it when Apple does it.
Perhaps because Google leaves their software in "beta" for decades. Google News was in beta for at least 3-4 years.
If I understand Siri correctly, a single Siri input starts with speech-to text (licensed from Nuance), then Siri's NLP (natural language processing) of the intended meaning, then converting that intended meaning to commands that are then made into specific queries outside of Siri that are then recreated in the most human readable form possible, or executed locally then displayed for you to confirm, and finally text-to-speech. All of these technologies have been around for decades, especially text-to-speech, but I can't think of one that ever been this accurate, useful or seamless. But it's all just obvious...
It's far from obvious; I didn't really understand what exactly Siri does until I watched Tom Gruber's presentation (I posted the link previously). Sure, the idea of what we want a virtual personal assistant to achieve is old, but the actual implementation of the language processing is novel, and has taken years of research.
Add to that that only a few years back we didn't have mobile processing power of today's magnitude, that GPS technology was not as widespread, that wireless internet connectivity was far from ubiquitous, even the internet itself isn't that old, and it only became conveniently accessible after Google's search engine... So no, it's not obvious at all -- at least it's not obvious how to achieve the effect, other than having a real assistant to do the things for you...
I think this is the first time Apple has put beta software front and center in an ad campaign. Maybe they are just trying to deflect any criticism of the lack of functionality by calling it beta. In fact I cannot recall any other instance when Apple released beta software publicly. This is clearly version 1.0.
When Google makes a general public software release and calls it beta, Apple fans joke how lame that is. But now, it seems they are ok with it when Apple does it.
Google doesn't sell its products, so nobody has any right to complain. If you don't like being a beta tester, don't use it...
I believe Siri's "beta" label comes not as much from the software not being of a "release" quality, but rather because for now it's rather USA-centric. I actually doubt that it will have matching features within a year in half of Europe, let alone the world. But then again, the market penetration of Apple will likely remain considerably higher in the US than elsewhere.
Siri is new! Sending all requests (an ever-increasing number) to the DC for analysis will allow Apple to quantify:
-- what type of requests are made most often
-- how those requests are worded
-- the followup request to verify the success of the initial request
It is likely, IMO, that Apple will try to concentrate on the high-volume/traffic requests by refining and tweaking the processing -- to the point where they can quickly recognize patterns (maybe voice prints) and phrasing that comprise these requests.
Then, they can start a parallel thread for the requests -- to test these refinements against the results of current processing.
This could be done on an aggregate level -- or individual requests tied to a specific device/user. I assume that the user would be asked to opt-in for the individual processing.
At some point, if there is a difference Siri could ask:
Siri: "here are the ways I understand what you requested: 1) ..... 2).... 3) neither"
The user would indicate the correct response.
Ultimately, for a select group of requests -- the detection and processing of these requests could be migrated back to the device as kind of a quick filter (or test) to see if the specific request could be resolved on the device, bypassing the server. Anything that fails the quick test is sent to the server, as normal
This would be an ongoing process -- as the devices become more powerful; Siri recognizes more requests and voice patterns; and Siri gains the ability to perform more actions.
This is almost like a parent-child or instructor-student relationship -- where, over time the knowledge/capability is passed down to where the student outperforms the teacher... so we just add a few more teachers and subjects... ad infinitum.
And the Siri servers already started buckle under the strain of 4 million iPhone 4S users in 3 days. These asshat trolls never think what would happen if Apple rolled out Siri to 120 million iDevices at once.
Good post. That pretty much sums up the situation.
Huh. Strange, then, that that's pretty much the reason for ALL of their success.
My first Mac, a IIsi, was downclocked intentionally so as not to compete with the IIfx. But there are many more examples of artificial product line differentiation. Of course there is nothing wrong with it per se, and yes you are right it's allowed massive profits for Apple. Most other computer companies do it, too.
But this thing with Siri just hurts. It's not the typical you lose a few Mhz from one computer to another. I would love to use it on my iPad 2.
However lots of people here have made a good point, it is in Beta. I imagine when they will eventually release it at least for iPhone 4 and iPad 2.
But this thing with Siri just hurts. It's not the typical you lose a few Mhz from one computer to another. I would love to use it on my iPad 2.
However lots of people here have made a good point, it is in Beta. I imagine when they will eventually release it at least for iPhone 4 and iPad 2.
That's what I believe, as well.
Heck, if they can release it for the iPhone 4, they can release it for the FIRST iPad, too. I'm not upset about it being limited to the 4S right now, but if they do decide to do a wider rollout and DON'T bring it to the original iPad, boy howdy, I'll be upset.
Heck, if they can release it for the iPhone 4, they can release it for the FIRST iPad, too. I'm not upset about it being limited to the 4S right now, but if they do decide to do a wider rollout and DON'T bring it to the original iPad, boy howdy, I'll be upset.
Me too, Buffalo Bob... and all you boyz and girlz in the peanut gallery!
Surprised the Apple apologists haven't come out yet, defending Apple's right to artificially restrict Siri to only the latest model, only to add a tiny bit more value to an otherwise underwhelming upgrade. "4 Steve," indeed. We all know this section would be full of insults and denigrations if Microsoft pulled this same crap with the latest Mango update.
1) Your's was the first comment -- so no apologists yet. Congratulations on first troll as well.
2) The 4S was of course and underwhelming upgrade -- it was intended to be just a MINOR upgrade with some costs shaved and a change in parts sourcing. If you hung out here more, you'd know this.
3) Putting their AI software on a particular upgraded device MIGHT have been to promote the sale - however, this being the first release of a groundbreaking new technology (granted, more of an implementation of various techs floating around for years), Apple MIGHT have wanted a limited release to work out the bugs and make sure it didn't BREAK.
If they had allowed it on the iPhone 3 -- and it garbled commands, you'd be here to point out how Lame the software is, and how they take credit for reintroducing other people's software done better elsewhere.
>> Otherwise, I'm sort of happy it has been hacked, so that I can test it out. Apple sometimes can annoy us with their need for profits.
All those open source peripherals I use with my XBox are great as well -- oh yeah, after 3 years in the wild I can network it without the $99 official device. High five anyone?
... lets' PLEASE not devolve into a tit-for-tat rant on closed or open or free vs profit. I think Apple has been fairly good at releasing things on various "Mac" platforms over time. But I appreciate them not hindering future products supporting the bast device forever. I can be ANNOYED Siri isn't out for my Mac.
... what I'd REALLY like is an emulator to run iPad/iPhone apps on my laptop because my kids never let me near the iPad since they got their hands on it.
>> Suffice to say; if Apple sucks so much -- why do you CARE that they don't have Siri on your device?
1) If you are locked into a contract why does it even matter?
2) The entitlement you people have is unreal. If Apple was trying to force you to update your current device by making it appear inferior why would they give you any update to the device, mucless going back three generations for the same basic OS. They'd just follow their competition's lead and never release a major update.
1) Because I'm locked into a contract is the reason why it matters. The problem isn't hardware related, it is software, which the upgrade is free.
2) I expect software to be the same for all the devices that are capable of running it. If Siri is limited to the US, it isn't about hardware, it is about creating a tiered customer base. The iPhone 4 is now inferior to the iPhone 4S because of the software.
I expect software to be the same for all the devices that are capable of running it.
Since Apple's Siri servers (I remind you, still in Beta) can't serve 120 million iOS-based at once your solution is to A) Not include Siri in any device for a gradual rollout or to include Siri in all devices even if means that no one can use it and Siri looks like a failure on a scale that towers over the MobileMe open release. I'm glad you don't work for Apple.
Quote:
The iPhone 4 is now inferior to the iPhone 4S because of the software.
Since Apple's Siri servers (I remind you, still in Beta) can't serve 120 million iOS-based at once your solution is to A) Not include Siri in any device for a gradual rollout or to include Siri in all devices even if means that no one can use it and Siri looks like a failure on a scale that towers over the MobileMe open release. I'm glad you don't work for Apple.
Are we forgetting something? Um, hardware?
Apple never releases beta software. This is highly unusual for Apple. Wait until the jailbreak and you'll see that it has nothing to do with hardware.
Comments
He's completely right though. We've had accurate voice recognition on phones for years.
Siri is only doing a few things differently. This includes:
-Contextual commands with a little more variety than the competition. Eg. "wake me in one hour"
-Some funny responses when you ask it weird questions
-Deep integration with system applications.
Other than that, Siri is nothing new. This should by no means be a selling point for anyone to buy a phone.
Ask it "How to tie a tie"
It returns with a google search. Just with a different interface. It will not actually tell you.
Aside from the commands that interact directly with the OS, that's all it really is. It's a Google search that presents the information neatly.
The voice technology is not new, or any more accurate than other current voice detection software. All the phone is doing is converting your voice to text. The text is sent off to Apple's servers to be processed, and for a command to be sent back. There are trigger words that define the context. Eg. "where is", "wake me up", "what's the weather", "play this album".
Is it currently the best voice command software on a phone? Yes.
Is this amazing new technology? No. But props to apple for integrating it deeply and making it easy and intuitive. But in the end, Siri is either controlling a limited amount of system apps, or it's doing a Google search and displaying the most relevant info based on your search
The difference is one of scale and opportunity.
When Siri is actively running on hundreds of millions of iDevices, and being monitored for accuracy, performance, unfulfillable requests (in multiple languages) -- it will likely generate a greater mine of information than all the prior efforts combined.
This will provide the opportunity to determine what is being done well, done badly or not done at all -- and too flesh out Siri to be even better.
If Apple exposes Siri's capabilities to 3rd-party apps and developers... we will have hundreds of thousands of talented, knowledgeable technical people testing and contributing to this rather specialized subset of AI.
I think that Siri is bigger than a single class of device, a predefined set of tasks -- or a single company.
Siri will become generic!
Yes, Exactly!
you: "Siri: text my wife I love you and I miss you"
you: "Siri: send it"
not
you: "Siri: open the messages app"
you: "Siri: select new message"
you: "Siri: find lucy"
you: "Siri: select lucy"
you: "Siri: skip the subject"
you: "I love you and I miss you"
you: "Siri: send it"
Ah, a much better example. I can see some fringe cases for how simply opening the app would be all that is needed, but I have a hard time imagining Apple releasing an API for a small group or users with a high chance the majority could abuse it.
It's funny how now we get people saying how Siri is nothing special, nothing new, nothing that wasn't already available and completely obvious.
If I understand Siri correctly, a single Siri input starts with speech-to text (licensed from Nuance), then Siri's NLP (natural language processing) of the intended meaning, then converting that intended meaning to commands that are then made into specific queries outside of Siri that are then recreated in the most human readable form possible, or executed locally then displayed for you to confirm, and finally text-to-speech. All of these technologies have been around for decades, especially text-to-speech, but I can't think of one that ever been this accurate, useful or seamless. But it's all just obvious...
Watch the video. It's a little creepy.
(via NasserAE)
All of those phones will be banned for life!
I haven't gotten through the comments, so it may have been stated already, although I'm not planning to wade through the fan boy hate.
It would seem to me this is a possibility. If Apple only wants this on 4S phones, my understanding was Siri goes out to the Apple DC. If that communication carries with it the model, it would seem easy enough to deny traffic from non-4S models and if it carried with ie the IMEI or other identifying info, it could disable the service or other Apple provided services. The latter may introduce privacy concerns and a legal battle with consumers it doesn't want to take up, but the first seems easily doable.
I'm not for it either way, but I believe it would be entirely possible and within their rights as the DC host to do so.
I think this is the first time Apple has put beta software front and center in an ad campaign. Maybe they are just trying to deflect any criticism of the lack of functionality by calling it beta. In fact I cannot recall any other instance when Apple released beta software publicly. This is clearly version 1.0.
When Google makes a general public software release and calls it beta, Apple fans joke how lame that is. But now, it seems they are ok with it when Apple does it.
Perhaps because Google leaves their software in "beta" for decades. Google News was in beta for at least 3-4 years.
...
If I understand Siri correctly, a single Siri input starts with speech-to text (licensed from Nuance), then Siri's NLP (natural language processing) of the intended meaning, then converting that intended meaning to commands that are then made into specific queries outside of Siri that are then recreated in the most human readable form possible, or executed locally then displayed for you to confirm, and finally text-to-speech. All of these technologies have been around for decades, especially text-to-speech, but I can't think of one that ever been this accurate, useful or seamless. But it's all just obvious...
It's far from obvious; I didn't really understand what exactly Siri does until I watched Tom Gruber's presentation (I posted the link previously). Sure, the idea of what we want a virtual personal assistant to achieve is old, but the actual implementation of the language processing is novel, and has taken years of research.
Add to that that only a few years back we didn't have mobile processing power of today's magnitude, that GPS technology was not as widespread, that wireless internet connectivity was far from ubiquitous, even the internet itself isn't that old, and it only became conveniently accessible after Google's search engine... So no, it's not obvious at all -- at least it's not obvious how to achieve the effect, other than having a real assistant to do the things for you...
I think this is the first time Apple has put beta software front and center in an ad campaign. Maybe they are just trying to deflect any criticism of the lack of functionality by calling it beta. In fact I cannot recall any other instance when Apple released beta software publicly. This is clearly version 1.0.
When Google makes a general public software release and calls it beta, Apple fans joke how lame that is. But now, it seems they are ok with it when Apple does it.
Google doesn't sell its products, so nobody has any right to complain. If you don't like being a beta tester, don't use it...
I believe Siri's "beta" label comes not as much from the software not being of a "release" quality, but rather because for now it's rather USA-centric. I actually doubt that it will have matching features within a year in half of Europe, let alone the world. But then again, the market penetration of Apple will likely remain considerably higher in the US than elsewhere.
Siri is new! Sending all requests (an ever-increasing number) to the DC for analysis will allow Apple to quantify:
-- what type of requests are made most often
-- how those requests are worded
-- the followup request to verify the success of the initial request
It is likely, IMO, that Apple will try to concentrate on the high-volume/traffic requests by refining and tweaking the processing -- to the point where they can quickly recognize patterns (maybe voice prints) and phrasing that comprise these requests.
Then, they can start a parallel thread for the requests -- to test these refinements against the results of current processing.
This could be done on an aggregate level -- or individual requests tied to a specific device/user. I assume that the user would be asked to opt-in for the individual processing.
At some point, if there is a difference Siri could ask:
Siri: "here are the ways I understand what you requested: 1) ..... 2).... 3) neither"
The user would indicate the correct response.
Ultimately, for a select group of requests -- the detection and processing of these requests could be migrated back to the device as kind of a quick filter (or test) to see if the specific request could be resolved on the device, bypassing the server. Anything that fails the quick test is sent to the server, as normal
This would be an ongoing process -- as the devices become more powerful; Siri recognizes more requests and voice patterns; and Siri gains the ability to perform more actions.
This is almost like a parent-child or instructor-student relationship -- where, over time the knowledge/capability is passed down to where the student outperforms the teacher... so we just add a few more teachers and subjects... ad infinitum.
And the Siri servers already started buckle under the strain of 4 million iPhone 4S users in 3 days. These asshat trolls never think what would happen if Apple rolled out Siri to 120 million iDevices at once.
Good post. That pretty much sums up the situation.
Watch the video. It's a little creepy.
VERY creepy but interesting.
Huh. Strange, then, that that's pretty much the reason for ALL of their success.
My first Mac, a IIsi, was downclocked intentionally so as not to compete with the IIfx. But there are many more examples of artificial product line differentiation. Of course there is nothing wrong with it per se, and yes you are right it's allowed massive profits for Apple. Most other computer companies do it, too.
But this thing with Siri just hurts. It's not the typical you lose a few Mhz from one computer to another. I would love to use it on my iPad 2.
However lots of people here have made a good point, it is in Beta. I imagine when they will eventually release it at least for iPhone 4 and iPad 2.
But this thing with Siri just hurts. It's not the typical you lose a few Mhz from one computer to another. I would love to use it on my iPad 2.
However lots of people here have made a good point, it is in Beta. I imagine when they will eventually release it at least for iPhone 4 and iPad 2.
That's what I believe, as well.
Heck, if they can release it for the iPhone 4, they can release it for the FIRST iPad, too. I'm not upset about it being limited to the 4S right now, but if they do decide to do a wider rollout and DON'T bring it to the original iPad, boy howdy, I'll be upset.
That's what I believe, as well.
Heck, if they can release it for the iPhone 4, they can release it for the FIRST iPad, too. I'm not upset about it being limited to the 4S right now, but if they do decide to do a wider rollout and DON'T bring it to the original iPad, boy howdy, I'll be upset.
Me too, Buffalo Bob... and all you boyz and girlz in the peanut gallery!
Surprised the Apple apologists haven't come out yet, defending Apple's right to artificially restrict Siri to only the latest model, only to add a tiny bit more value to an otherwise underwhelming upgrade. "4 Steve," indeed. We all know this section would be full of insults and denigrations if Microsoft pulled this same crap with the latest Mango update.
1) Your's was the first comment -- so no apologists yet. Congratulations on first troll as well.
2) The 4S was of course and underwhelming upgrade -- it was intended to be just a MINOR upgrade with some costs shaved and a change in parts sourcing. If you hung out here more, you'd know this.
3) Putting their AI software on a particular upgraded device MIGHT have been to promote the sale - however, this being the first release of a groundbreaking new technology (granted, more of an implementation of various techs floating around for years), Apple MIGHT have wanted a limited release to work out the bugs and make sure it didn't BREAK.
If they had allowed it on the iPhone 3 -- and it garbled commands, you'd be here to point out how Lame the software is, and how they take credit for reintroducing other people's software done better elsewhere.
>> Otherwise, I'm sort of happy it has been hacked, so that I can test it out. Apple sometimes can annoy us with their need for profits.
All those open source peripherals I use with my XBox are great as well -- oh yeah, after 3 years in the wild I can network it without the $99 official device. High five anyone?
... lets' PLEASE not devolve into a tit-for-tat rant on closed or open or free vs profit. I think Apple has been fairly good at releasing things on various "Mac" platforms over time. But I appreciate them not hindering future products supporting the bast device forever. I can be ANNOYED Siri isn't out for my Mac.
... what I'd REALLY like is an emulator to run iPad/iPhone apps on my laptop because my kids never let me near the iPad since they got their hands on it.
>> Suffice to say; if Apple sucks so much -- why do you CARE that they don't have Siri on your device?
1) If you are locked into a contract why does it even matter?
2) The entitlement you people have is unreal. If Apple was trying to force you to update your current device by making it appear inferior why would they give you any update to the device, mucless going back three generations for the same basic OS. They'd just follow their competition's lead and never release a major update.
1) Because I'm locked into a contract is the reason why it matters. The problem isn't hardware related, it is software, which the upgrade is free.
2) I expect software to be the same for all the devices that are capable of running it. If Siri is limited to the US, it isn't about hardware, it is about creating a tiered customer base. The iPhone 4 is now inferior to the iPhone 4S because of the software.
I expect software to be the same for all the devices that are capable of running it.
Since Apple's Siri servers (I remind you, still in Beta) can't serve 120 million iOS-based at once your solution is to A) Not include Siri in any device for a gradual rollout or
The iPhone 4 is now inferior to the iPhone 4S because of the software.
Are we forgetting something? Um, hardware?
Since Apple's Siri servers (I remind you, still in Beta) can't serve 120 million iOS-based at once your solution is to A) Not include Siri in any device for a gradual rollout or
Are we forgetting something? Um, hardware?
Apple never releases beta software. This is highly unusual for Apple. Wait until the jailbreak and you'll see that it has nothing to do with hardware.
Apple never releases beta software.
Abject lie?
Apple never releases beta software. This is highly unusual for Apple. Wait until the jailbreak and you'll see that it has nothing to do with hardware.
Fact: Apple release Betas all the time.
Fact: The iPhone 4S is superior to the iPhone 4 in HW.
Fact: Apple release Betas all the time.
Fact: The iPhone 4S is superior to the iPhone 4 in HW.
1. Really? List three major betas released to the public? (I'll even give you one, Facetime).
2. Duh. Who would have thought that the 4S wouldn't have superior hardware? However, this upgraded hardware is not needed for Siri.
Just sayin'.