PDA

View Full Version : Voice, Speech, and/or sound recognition


mweed
07-11-2007, 02:06 PM
Since the topic of Pleo's ability (or lack thereof) in voice and speech recognition appears in over 14 of the 64 existing threads, I think we need to consolidate those discussions into their own thread.

First off, according to PleoWorldHostCami:
"Pleo Does NOT have voice recognition, so he will not "learn" a name or specific voice commands." :(

And according to Alex Godden:
Just to be clear, Pleo does not have voice recognition capability and this was never planned, but he can hear and be trained by sensory input such as feeding.

He responds to noises and will turn to look at them, but he cannot learn his name, any language or specific commands. There may be other capabilities possible in the future, but nothing else is confirmed yet.

And, a definition by roschler:

Voice recogntion is the art of identifying one person from another by the sound or tonal characteristics of their voice.

Speech recognition is the art of converting speech to text, whether for dictation or "command and control" purposes.

The big questions from here are: what "other" capability does Pleo have, and can we build on that capability?

mweed
07-11-2007, 02:10 PM
Speech recognition akin to Dragonsoft's "Naturally Speaking" software with a large vocabulary is a major undertaking well beyond what anyone would expect. But what I would call "sound recognition" should be several orders of magnitude easier. That is, getting a pleo to respond to specific sound such as a double-clap, a specific whistle tone, or single syllable word like "no", "bark", or "beg" should be doable. And, like a lot of speech recognition software, the speech sample used by the speech engine should be "learned" by the pleo as it is spoken by the owner (the person "training" the pleo).

I admit it's an oversimplification, but speech recognition with reasonably large vocabularies was in the 70-80% accurate range running back in the Pentium II days. So, I would think a severely limited vocabularywith even a iffy accuracy should be possible. If a pleo responded to a 5 sound vocabulary 70% of the time, most people would be ecstatic, and assume the other 30% of the time the pleo was simply ignoring them.

That would tend to inspire people to come up with simple one or two syllable names like Bob instead of Benedictine. But hey, it's a start.

amjoie
07-11-2007, 03:08 PM
I totally agree this thread was needed. LOL

What we DO know is that Pleo responds to tone. Harsh/angry tone = sad/depressed and pleasant/sweet tone = happy.

So a "no" said in a harsh tone is going to have a negative impact, while a "good boy/girl" said in a happy tone is going to be a positive reinforcement.

What we don't yet know is IF a negative response will result in Pleo staying away from a certain activity; or does a negative response just make him get sad (or maybe even angry) and need a cheering-up session.

Could a PleoHost find out if Pleo eventually avoids behavior that is followed by a harsh tone? In other words, *regarding the arena of training* does Pleo respond to negative reinforcement, or only to positive reinforcement?

alibaba
07-11-2007, 03:41 PM
What we DO know is that Pleo responds to tone. Harsh/angry tone = sad/depressed and pleasant/sweet tone = happy.

How do we know this? It isn't in the FAQs and I can't find it anywhere on the PleoWorld site.

The site is pretty vague about specific reactions to specific stimuli, I'm guessing because Pleos will respond differently depending on their 'upbringing' - which is why they are so lifelike!

AB

alibaba
07-11-2007, 03:45 PM
Sorry, replying to my own post which I know is not polite!

How could a Pleo reliably tell the difference between a harsh tone and just a deeper voice? Are you thinking he will 'learn' his owner's voice? Or will Pleos just get upset when people with gruff voices talk to them, even if they mean to be nice? Is that sexist ;)?

AB

amjoie
07-11-2007, 04:09 PM
How do we know this? It isn't in the FAQs and I can't find it anywhere on the PleoWorld site.

The site is pretty vague about specific reactions to specific stimuli, I'm guessing because Pleos will respond differently depending on their 'upbringing' - which is why they are so lifelike!

AB

I've been following Pleo's development for a while, and read almost everything I can find. Somewhere in all that info, and somehow when metabolizing all of it, my brain came to this conclusion.

But, yes, you are right. Until we have it verified on this forum, a conclusion should not be drawn.

So, PleoHosts, is my statement correct, or is it now outdated info, or did my brain take a wrong turn and merely assume?

PleoWorldHostSauri
07-11-2007, 05:48 PM
Great points alibaba. Amjoie I will see what I can find out about how Pleo's "hearing" works and how he processes what he "hears.":)

mweed
07-12-2007, 09:31 AM
So a "no" said in a harsh tone is going to have a negative impact, while a "good boy/girl" said in a happy tone is going to be a positive reinforcement.

I guess we need to add "tone recognition" to the list of voice, speech and sound in this thread... ;)

I know when it comes to dogs, tone is everything. I can yell "Good dog!" in an angry tone and the dog acts hurt. I also quite frequently say things in a sweet, loving tone like "You're a worthless mutt. It wouldn't be worth the cost of the bullet to shoot you." and the dog gets all happy and excited, wagging its tail... :rolleyes: Absolutely no speech recognition there. And it's not the recognition of my voice or of a specific sound/word. It's purely tone.

alibaba
07-12-2007, 11:11 AM
And probably expression/body language too - it's very hard to say something in a sweet, loving voice while looking angry. Conversely, if you're making your voice sound harsh you'll probably find yourself frowning, clenching your fists etc.

I'm pretty sure Pleo won't be advanced enough to notice that, but I'm sure your dog does!

AB

amjoie
07-12-2007, 01:18 PM
And probably expression/body language too - it's very hard to say something in a sweet, loving voice while looking angry. Conversely, if you're making your voice sound harsh you'll probably find yourself frowning, clenching your fists etc.

I'm pretty sure Pleo won't be advanced enough to notice that, but I'm sure your dog does!

AB

Dogs, cats, horses, and probably lots of other living critters, can sense things people cannot sense: fear, mood, positive or negative energy, tension energy vs relaxed energy, dominance energy vs submission energy, a heightened sense of smell and hearing, etc.

Some of them do understand a small number of actual words, but most give much more weight to their other senses because those senses cannot lie.

One of the problems with robots is that they have no sensing abilities beyond what we give them. And since we do not fully know or understand the mechanisms that allow living creatures to sense things like fear or various forms of energies, we cannot give the robots those abilities.

And some senses, like smell, are beyond our ability to give to a robot, even though we know a lot about how smell operates.

That means interaction with robots will be strictly based on a very limited number of finite sensing capabilities -- not nearly enough of them to satisfy the perception that the robot is actively "on our wave-length" like living pets.

For instance, a single command, like "come" does not signal automatic obedience in a living creature. It is more the attention getting device that makes the animal focus on the human. Once focused, the animal has to "get the message" by figuring out what the human actually wants.

So if the word "come" is accompanied by a change in posture and facial expression, plus a change in energy (and perhaps even sound and smell), then the living creature is drawn toward the person and rewarded for approaching. With repeated trial and error, the living creature finally realizes the sound of the command will always mean the same thing -- and they are trained to the "come" command.

But they haven't just learned a word. They have learned a series of events/senses that mean something when attached to the command. And if the command is significantly changed in tone, energy, or non-verbal expression, they will relearn the command to mean something entirely different. So, if the command to "come" at one time meant "approach quickly," it could easily change to mean "run away quickly because you are about to be scolded/bathed/taken-to-the-vet."

But if we program a robot to "come" with voice recognition, it will always mean the same exact thing. The robot will have no choice but to approach (absent the programming to sometimes ignore a command, or the inability at certain times to accurately hear us).

If, however, we program the robot to learn what "come" means, then we must give the same exact communication each time we utter the command, in order for the robot to sense our command word. And it must be a non-variable communication the robot understands, repeated over and over until learning has taken place.

Training on this level is very difficult for humans to accomplish, because we vary everything, depending on our mood or physical condition. A person with a cold does not sound the same as a person's normal voice. An angry, frightened, tense or frustrated person does not act or sound the same as a person in a normal situation. How much pressure we place on an object depends on mood or physical condition. Nothing is black and white -- everything is shades of gray with humans.

So training robots is a challenge. Much more so, than training living creatures.

I am looking forward to seeing how the people at Ugobe have managed to overcome the obstacles involved in robot AI training, on a Pleo level of sensing ability.

Regarding the sensing ability, I am just guessing here, but this is what I expect Pleo will have for sensing ability -- if I am right, Pleo will be able to sense the following: abrupt changes in somewhat exaggerated tone; a certain level of pressure on his sensing pads; abrupt noises; the proximity of IR in another life form; closeness to a vertical surface; too much tension on his joints; some color as well as light/shadow.

Under good lighting conditions, I am guessing Pleo will also sense large differences in color, and abrupt changes in shadow vs light (helping him to judge whether he is on level ground vs a drop-off).

IF these guesses are correct, you can see our ability to communicate with Pleo will be very limited. Sound is probably not as important, alone, as the combination of sound and touch. Sight is probably his least active sense, regarding communication.

So I am guessing that Pleo will need close and rather constant attention to learn from us. But I'm sure he will also learn things by interacting with his environment and with other Pleos.

I think the field of robot "sensations" is developing rapidly, and I look forward to great changes in the near future. But until then, we *shouldn't expect too much* from our little friend. We should just love him for what he is today -- a relatively low budget, but ground-breaking robotic life form.

mweed
07-12-2007, 03:37 PM
Dogs, cats, horses, and probably lots of other living critters, can sense things people cannot sense: fear, mood, positive or negative energy, tension energy vs relaxed energy, dominance energy vs submission energy, a heightened sense of smell and hearing, etc.

You are absolutely correct is saying that animals take advantage of all of these inputs as they are available, but most animals can do quite well with just one. There's truth in the old joke of cats coming running from the far side of the house at the sound of the can opener. Most dog trainers will teach you to use both verbal and visual cues; you say "come" and you wave your arm, or you point your finger and say "Bang!" But on the other hand, I know from experience that when our horse was on the far side of the pasture, quite a distance over a hill, out of visual and olfactory range, all I had to do was rattle her feed bucket and she'd come running. Her only cue was the audible sound. Once she got close, and could see and smell and sense, she might change her mind and trot off based on the additional inputs...


If, however, we program the robot to learn what "come" means, then we must give the same exact communication each time we utter the command, in order for the robot to sense our command word. And it must be a non-variable communication the robot understands, repeated over and over until learning has taken place.

Training on this level is very difficult for humans to accomplish, because we vary everything, depending on our mood or physical condition. A person with a cold does not sound the same as a person's normal voice. An angry, frightened, tense or frustrated person does not act or sound the same as a person in a normal situation. How much pressure we place on an object depends on mood or physical condition. Nothing is black and white -- everything is shades of gray with humans.


Consistency is the key when "training" for speech recognition, but it doesn't take that long with most programs I've used. I realize our pleos won't have the level of sophistication as the latest speech recognition software, but the capability is here, and getting better all the time. The latest version of Naturally Speaking does not require any training, and is 98%+ accurate out of the box.

It would be nice if someone from Ugobe could enlighten us on things like CPU speed, memory, etc. to help us determine what level of capability is there for us to experiment with...

ryane
07-12-2007, 04:12 PM
Pleo is a high budget toy, but a low budget "robot".

something that i have thought about a bit is that a robot can be made with more senses than humankind. they could be made to feel the colors they see , hear all frequencies, see invisible light (IR), (which they do already!), and basically experience the entire electromagnetic spectrum! eventually in our near future we could become obsolete, our senses made minimal by what we give our creations. this is inevitable.

and has anyone heard about the singularity? don't respond here, it would belong in another thread! just look it up.

amjoie
07-12-2007, 04:14 PM
Consistency is the key when "training" for speech recognition, but it doesn't take that long with most programs I've used. I realize our pleos won't have the level of sophistication as the latest speech recognition software, but the capability is here, and getting better all the time. The latest version of Naturally Speaking does not require any training, and is 98%+ accurate out of the box.

It would be nice if someone from Ugobe could enlighten us on things like CPU speed, memory, etc. to help us determine what level of capability is there for us to experiment with...

What effect do you think distance and ambient room noise will have on the ability to gain accurate voice recognition in a Pleo-type robot of the future?

Even with the advances in software, don't you have to be fairly close to the microphone, in a quiet room, for a good voice recognition software to work properly, with that degree of accuracy? I imagine that most places Pleo will play probably have bad acoustics and are fairly noisy.

And I wonder what effect Pleo's own joint-movement sounds will have on him being able to properly differentiate sound, on a voice recognition level.

I am guessing that to mimic the accuracy of a stationary desktop computer set up, in a moving life form, we will need much more advanced and sophisticated technology than what is needed for the desktop.

PleoWorldHostSauri
07-12-2007, 04:14 PM
It would be nice if someone from Ugobe could enlighten us on things like CPU speed, memory, etc. to help us determine what level of capability is there for us to experiment with...

Please be patient :) when the questions are technical, we have to get ahold of the development team, the guys who are busy building your Pleos ;)

Pirka
07-12-2007, 04:25 PM
I think the field of robot "sensations" is developing rapidly, and I look forward to great changes in the near future. But until then, we *shouldn't expect too much* from our little friend. We should just love him for what he is today -- a relatively low budget, but ground-breaking robotic life form.


Again, you should be a motivational speaker - you hit the nail on the head everytime with your long posts! Viva la amjoie - who will definately be a Pleo's best friend in the future. Or two as I have heard. ;)

Regarding the subject of this thread, I actually do agree with you all when it comes to direct voice recognition - in a way, not being able to know words is better, as long as he can 'hear', I'm ok! The fact he can supposedly hear tone makes it even better and more realistic.
I know when Pleo's in my arms, I won't care about all this - he'll just be the best. :D

- Pirka

ryane
07-12-2007, 04:28 PM
Please be patient :) when the questions are technical, we have to get ahold of the development team, the guys who are busy building your Pleos ;)

that's ok as long as they stay busy! no breaks faster, faster work like you mean it!

Thanks Ugobe, we know you share the same dream.

amjoie
07-12-2007, 04:46 PM
Pleo is a high budget toy, but a low budget "robot".

something that i have thought about a bit is that a robot can be made with more senses than humankind. they could be made to feel the colors they see , hear all frequencies, see invisible light (IR), (which they do already!), and basically experience the entire electromagnetic spectrum! eventually in our near future we could become obsolete, our senses made minimal by what we give our creations. this is inevitable.

and has anyone heard about the singularity? don't respond here, it would belong in another thread! just look it up.

Well, logistically, it is quite difficult to put all those sensors in a confined space, inside a moving life form. We already have sensors individually designed to do what you describe, but consider the equipment that houses them, individually. Imagine the equipment needed to combine them. It would take a great deal of miniaturization to make it feasible, and the cost would be through the roof. Nano technology does not come cheap.

I think the only answer, eventually, will be a combination of mechanics and biology in order to make things cost effective. But then you have to deal with mutations. Not pretty!

And having the ability to sense the whole electromagnetic spectrum is only half the battle. Even once the sensors are in place, programming is needed to interpret all that data. We have brains designed to interpret only a portion of the data, because all of it would be overwhelming to us.

No computer even begins to match the full capability of our brains. So your future bots would need an organ much, much more complex than our brain, to compute the data -- and programming to organize and interpret it. Talk about sensory overload!

And we haven't even touched on what kind of programming would be necessary to make any kind of tangible use of the interpreted data, in combination. Or the ability to use it wisely.

I think it is much more likely we will be limited by finances; and we will use future miniaturization capabilities to make smaller tools for our use, and pets for our enjoyment -- rather than fully capable creatures who could challenge our existence, and cost more than an entire country's annual budget.

ryane
07-12-2007, 05:02 PM
"the number of transistors on a chip will double every two years. Information technologies are doubling in power every year right now. Doubling every year is multiplying by 1,000 in ten years. It's remarkable how scientists miss this basic trend." By 2027, computers will surpass humans in intelligence; by 2045 or so, we will reach the Singularity."
-Kurzweil

amjoie
07-12-2007, 05:30 PM
"the number of transistors on a chip will double every two years. Information technologies are doubling in power every year right now. Doubling every year is multiplying by 1,000 in ten years. It's remarkable how scientists miss this basic trend." By 2027, computers will surpass humans in intelligence; by 2045 or so, we will reach the Singularity." -Kurzweil

In a perfect world.

With no world wars. No financial restraints. No megalomaniacs. No collapsing civilizations. No premature ice age. No reversal of magnetic poles. Etc.

Humans are infinitely capable. But they rarely reach their capabilities, and then only in a limited scope, and in small spurts.

However, I will most likely be long gone by 2045, so what do I care .... ;)

In the meantime, I'll settle for really exciting advances in small life forms. LOL

ryane
07-12-2007, 05:42 PM
"However, I will most likely be long gone by 2045, so what do I care"

i will still remain, i will only be turning 69 that year. if i live that long of course, but most people live into their eighties or older in this age.;)

amjoie
07-13-2007, 08:50 AM
"However, I will most likely be long gone by 2045, so what do I care"

i will still remain, i will only be turning 69 that year. if i live that long of course, but most people live into their eighties or older in this age.;)

If someone had told me, when I was your age, that I would be waiting for my own personal computerized AI life form to arrive in the fall of 2007, I would have scratched my head wondering what on earth the madman was talking about.

Because it was before Bill Gates and Steve Jobs made PCs for the masses, AI was not a household word, and robots were only the stuff of dreams. The notion of a Pleo, actually existing outside of science fiction, would have been unconceivable, incomprehensible.

If the old world is still turning, and a precarious worldwide political balance still exists, the things that make your everyday life worthwhile and fulfilling in 2045 have probably not even been invented, yet. Or if they have, only a few visionaries know about it, because today all that exists is a tiny piece of potential, which someone still needs to figure out how to apply and market.

And that invention will likely make everything we know about computer technology become nothing but antiquated history. Today's cutting edge technology will join hi-fi stereos playing vinyl records with a needle, and eight-track cassettes -- in the junk pile of what might have been -- because something so revolutionary was discovered that it took the world by storm, and buried everything before it.

For all we know, your household-appliance/servant-companion of the future might be a shape-shifting hunk of cosmic goo, controlled by your thoughts alone. Don't laugh, because stranger things have happened.

Fifty years is forever. You can't begin to see that far ahead. And when you get there, and look back at today, you will laugh, remembering the ill-fated forecasts of the future. (The way I laugh at the forward-thinkers in the 1950s who believed flying cars would be common well before the year 2000 -- something still not possible -- but they totally missed even the possibility of the internet and websites.)

From my vantage point, every epoch in history has its very own singularity -- then they recover, and move on. LOL

So I'll just look forward as far as October, and the arrival of Pleo. Maybe I'll stretch my vision to include Pleo's relatives, of the near future. If I look much further, I might be staring into eternity and beyond. You never know. ;)

mweed
07-13-2007, 09:00 AM
However, I will most likely be long gone by 2045, so what do I care .... ;)

Kurzweil plans to still be here. His idea is that not only is technology advancing, but so are all the other related fields, like medicine. We have already succeeded is extending life expectancy by 15+ years, and in another 10 years we will see even more advances, which would extend the ability to keep people going even longer. So, his view is that if current technology and medicine can keep you around for the next twenty years, by then technology will be 1,000,000 times further along, and we should have developed ways to keep people alive forever...

alibaba
07-13-2007, 10:03 AM
And we haven't even touched on what kind of programming would be necessary to make any kind of tangible use of the interpreted data, in combination. <b>Or the ability to use it wisely.</b>

(My emphasis)
I don't think even humans have got that part sorted out yet...

;)

AB

amjoie
07-13-2007, 10:29 AM
Kurzweil plans to still be here. His idea is that not only is technology advancing, but so are all the other related fields, like medicine. We have already succeeded is extending life expectancy by 15+ years, and in another 10 years we will see even more advances, which would extend the ability to keep people going even longer. So, his view is that if current technology and medicine can keep you around for the next twenty years, by then technology will be 1,000,000 times further along, and we should have developed ways to keep people alive forever...

ROTFLOL!!!

Tell anyone who spends much time in a hospital that medicine is advancing ... the more time you spend under today's medical care, the less you trust medicine. I won't relate the heartbreaking stories. The young don't want to know.

But just think about this for a minute: The governments of the world are doing everything they can do to stop overpopulation. Do you really believe they are going to invest in ways to make humans immortal? They want you to work hard, pay your taxes, fight their wars, have 1.2 children, and then die before you need expensive medical attention.

I love Utopian dreams. I enjoy reading about them, and I love theorizing about and discussing them, as if they had merit. But I also know the difference between dreams and reality. Death cannot be escaped. All men, in final humility, bend their knee one last time. Even the stars in the sky come to an end ....

But behold the optimist in me: I believe little Pleo will have batteries in the near future with a greater charge capacity, and longer life. I believe he will have better hearing, and better sight. I believe his touch sensors will eventually cover more of his body, and be more sensitive. I believe Pleo will get smarter and be able to do more. I believe lay programmers will add to Pleo's abilities and to my enjoyment of him. I believe Pleo will be only one member of a greater life form family. So, the future looks bright. :D

Don't look too far ahead into the future, except for "kicks and giggles." Otherwise just live each day to its fullest. You'll live longer. ;)

mweed
07-13-2007, 12:07 PM
But behold the optimist in me: I believe little Pleo will have batteries in the near future with a greater charge capacity, and longer life. I believe he will have better hearing, and better sight. I believe his touch sensors will eventually cover more of his body, and be more sensitive. I believe Pleo will get smarter and be able to do more. I believe lay programmers will add to Pleo's abilities and to my enjoyment of him. I believe Pleo will be only one member of a greater life form family. So, the future looks bright. :D

My hope is that the designers at Ugobe planned well enough on the hardware that as their LifeOS is improved over the next few years, we'll be able to load the much more sophisticated versions of LifeOS into our original Pleos, rather than having to buy newer versions. That way, even though Pleo does not have speech recognition today, he might in a year or two...

And on the funny side: according to the Fortune magazine article on Kurzweil, he take over 200 vitamin/supplements a day to keep his body young... :eek:

mweed
07-14-2007, 10:21 PM
I noticed in the Wikipedia listing for pleo that he has a 32-bit NXP ARM 7 sub processor specifically dedicated to the camera system and audio input. So I hope that means someday he'll be able to master speech recognition, it will just take time and talented programmers...

roschler
07-15-2007, 04:35 AM
ROTFLOL!!!

Tell anyone who spends much time in a hospital that medicine is advancing ... the more time you spend under today's medical care, the less you trust medicine. I won't relate the heartbreaking stories. The young don't want to know.

But just think about this for a minute: The governments of the world are doing everything they can do to stop overpopulation. Do you really believe they are going to invest in ways to make humans immortal? They want you to work hard, pay your taxes, fight their wars, have 1.2 children, and then die before you need expensive medical attention.

I love Utopian dreams. I enjoy reading about them, and I love theorizing about and discussing them, as if they had merit. But I also know the difference between dreams and reality. Death cannot be escaped. All men, in final humility, bend their knee one last time. Even the stars in the sky come to an end ....

But behold the optimist in me: I believe little Pleo will have batteries in the near future with a greater charge capacity, and longer life. I believe he will have better hearing, and better sight. I believe his touch sensors will eventually cover more of his body, and be more sensitive. I believe Pleo will get smarter and be able to do more. I believe lay programmers will add to Pleo's abilities and to my enjoyment of him. I believe Pleo will be only one member of a greater life form family. So, the future looks bright. :D

Don't look too far ahead into the future, except for "kicks and giggles." Otherwise just live each day to its fullest. You'll live longer. ;)

I am writing a program that will transfer my consciousness into my Pleo. Then I will be immortal and people will pet me and think I am cute for all eternity. :D

ryane
07-15-2007, 04:48 AM
I am writing a program that will transfer my consciousness into my Pleo. Then I will be immortal and people will pet me and think I am cute for all eternity. :D

awesome! me too!

mweed
07-19-2007, 07:57 PM
Here's a quote from Alex Godden in another thread:

There are no plans for UGOBE to provide voice/speech recognition, although Pleo's ‘ears’ (microphones) feed into 10-bit A/D converters in a 32-bit ARM7 so theoretically the possibility exists for such a capability in the future, given the right software program.

Alex
UGOBE

pleoguy101
07-20-2007, 07:34 AM
I noticed in the Wikipedia listing for pleo that he has a 32-bit NXP ARM 7 sub processor specifically dedicated to the camera system and audio input. So I hope that means someday he'll be able to master speech recognition, it will just take time and talented programmers...

And i hope someday is before october!!:)

pleoguy101
07-20-2007, 07:46 AM
My hope is that the designers at Ugobe planned well enough on the hardware that as their LifeOS is improved over the next few years, we'll be able to load the much more sophisticated versions of LifeOS into our original Pleos, rather than having to buy newer versions. That way, even though Pleo does not have speech recognition today, he might in a year or two...

And on the funny side: according to the Fortune magazine article on Kurzweil, he take over 200 vitamin/supplements a day to keep his body young... :eek:

i agree with the voice recognition being downloaded into our pleo, but hopefully pleo will have voice recognition before october!!!:D :)

amjoie
07-21-2007, 03:12 PM
There are no plans for UGOBE to provide voice/speech recognition, although Pleo's ‘ears’ (microphones) feed into 10-bit A/D converters in a 32-bit ARM7 so theoretically the possibility exists for such a capability in the future, given the right software program.

Alex
UGOBE

I am writing a program that will transfer my consciousness into my Pleo. Then I will be immortal and people will pet me and think I am cute for all eternity. :D

Well, roschler, once you are inside Pleo, you can write a new voice/speech recognition program, too, taking more full advantage of the under-used Pleo hardware. That way, people can converse with you, and love you for your mind, and not just your body. LOL

ryane
07-21-2007, 03:56 PM
amjoie!

i didn't think you could stay away!

welcome!

amjoie
07-21-2007, 05:50 PM
amjoie!

i didn't think you could stay away!

welcome!

Thank you! I *did* say I'd be back ... :)

I'll be in and out, sometimes posting and sometimes lurking. I don't want to be totally out of commission again, so I have to be a good little girl. Or try to be, anyway. LOL

cmadmacs
07-21-2007, 06:07 PM
welcome back home.

Brokenh3art
09-16-2007, 04:23 PM
ive already preordered a pleo but im about to preorder a second one, the first one will be my expirment where i will descramble the windows vista voice recognitions code and attempt to input it into pleo, if it can be done / or even if the developers say it isnt possible i am going to try.

in hopes of one day making my pleo come to me with the command of a simple toon "come here pleo"

and then when my family it pleo wouldnt recognize and wouldnt listen.

that way pleo would be truley "Unique"

gimme youre input pleoworldhostcami since youve been very helpfull id like to see what thoughts are,

And everyone else who dreams of voice recognition gimme youre input and device
and a sorry english type skill not soo good

PleoWorldHostCami
09-16-2007, 06:10 PM
ive already preordered a pleo but im about to preorder a second one, the first one will be my expirment where i will descramble the windows vista voice recognitions code and attempt to input it into pleo, if it can be done / or even if the developers say it isnt possible i am going to try.

in hopes of one day making my pleo come to me with the command of a simple toon "come here pleo"

And everyone else who dreams of voice recognition gimme youre input and device
and a sorry english type skill not soo good

There are no plans for UGOBE to provide voice/speech recognition, although Pleo's ‘ears’ (microphones) feed into 10-bit A/D converters in a 32-bit ARM7 so theoretically the possibility exists for such a capability in the future, given the right software program.

On our upcoming developers' page we will be sharing the technical details necessary for independent software programmers and developers to write programs giving Pleo more advanced capabilities like these, which we hope they will share with other Pleo owners.

mweed
09-16-2007, 07:26 PM
My cheap cell phone has name recognition for hands-free dialing. And that's done in i Java ona much less powerful CPU, so I believe it can be done.

cmadmacs
09-23-2007, 04:36 PM
We did put a man on the moon. So anything is possible.

averyiscool
12-30-2007, 03:40 PM
can pleo be trained to come when you call him? if he can leave a comment.:confused:

Lu Bu
12-30-2007, 05:55 PM
Some people claimed to say he can. I did, however, see a vid where a kid called his Pleo and his Pleo ambled towards him. I assumed that he trained his Pleo to do so. I can tell you how to train your Pleo how to do that:

1. Make sure your Pleo is interacting with you
2. Hold Pleo's leg for about 3-4 seonds
3. Wait till he gets to you
4. Once he walks torwards you, praise him so that he can be encouraged to do so (I am not sure if this feature(4) will work since Pleo's current software is only made for subtle personality shaping.)

Pleo_Joe
12-31-2007, 11:05 AM
can pleo be trained to come when you call him? if he can leave a comment.:confused:

Although Pleo can hear sounds, I don't believe there is any evidence that Pleo will actually walk to you when you call it.
I have several videos in my blog that show I cannot make my Pleo walk to me on command.
The only thing that makes it look good is to be in front of your Pleo, calling to it, when it starts walking. It's definitely a fun thing to do, but it is not having Pleo come when called.
If you face the Pleo away from you and call to it, it may (or may not) turn to look at you, but it's more than likely NOT going to turn around and walk to you.

If you're looking to get a Pleo, don't let this reply dissuade you - I really like my Pleo and continue to have fun with him every day! :)

Lu Bu
01-02-2008, 07:51 PM
Pleo's microphones were designed to look for the sound's source, not recognize speech commands. I assume because of this, Pleo can walk towards the sound/voice, though this does not count as a trick.

Pleo_Joe
01-02-2008, 07:59 PM
Pleo's microphones were designed to look for the sound's source, not recognize speech commands. I assume because of this, Pleo can walk towards the sound/voice, though this does not count as a trick.

I agree with you, Lu Bu. To be specific, however, I would say that Pleo *CAN* walk towards the sound/voice, but I do not believe it to be in the current programming to actually DO it.

Lu Bu
01-02-2008, 08:03 PM
I agree with you, Lu Bu. To be specific, however, I would say that Pleo *CAN* walk towards the sound/voice, but I do not believe it to be in the current programming to actually DO it.

Precisley, I agree! :)

eric
01-10-2008, 12:42 PM
Tyler

Could pleo be updated to add voice commands ?


Technically i mean ?

ryane
01-10-2008, 01:30 PM
Tyler

Could pleo be updated to add voice commands ?


Technically i mean ?

that's a good question for Tyler...:confused:

Lu Bu
01-13-2008, 06:15 PM
It depends. Pleo's current microphones can be upgraded to have voice recognition, but I am not sure if he will be able to have speech recognition via software. BUT with the PLEO SDK it might be possible, but very complex and really hard.

PleoPet
01-13-2008, 07:12 PM
Pleo's sound localization and camera and IR and a few other things are handled by a separate CPU.
The second smaller ARM processor, 32KB of Flash ROM for programs, 8KB of RAM -- 1/8th the memory of the main brain. The two CPUs are connected, but through a slow serial link.
http://www.aibohack.com/pleo/tech1.htm
http://forums.pleoworld.com/showthread.php?t=1167

There isn't much room for new firmware, but it is possible. You may need to reduce the camera features to add more sound features.

If UGOBE would release source and hardware specs for the 2nd CPU, I suspect a few people would try to implement something like this, and other audio and camera features. Simpler speaker independent voice commands should be possible.

BTW: The Pawn SDK doesn't address this kind of problem. Much more difficult than writing a simple Pawn script.

....
> Genibo = Vaporware Robot..., BEST OF ALL ... you decide

Lu Bu
01-15-2008, 08:08 AM
Dang! So if I wanted to add more sound features, I'd have to rduce the camera? Too bad the 2nd processor ain't as powerful as the first.

chamaeleo
01-17-2008, 08:32 AM
Hahaha! I am sooo glad I live with Pleohacker.

pleoWhisperer
02-08-2008, 03:40 PM
If hasbro can do it fro 1/2 the price of pleo . . .

their new robotic golden retriever puppy claims recognition of six phrases

http://money.cnn.com/galleries/2008/news/0802/gallery.toy_fair/7.html

milentije
02-08-2008, 03:52 PM
Disney did it several years ago with their "Lelo & Stitch" doll. I have him, and he obeys or answers several different commands and questions. But i do agree. It is the only thing that he needs to be the best thing ever!

pleoguy101
02-08-2008, 07:18 PM
Disney did it several years ago with their "Lelo & Stitch" doll. I have him, and he obeys or answers several different commands and questions. But i do agree. It is the only thing that he needs to be the best thing ever!

I have that too.. but just because it knows some certain phrases doesn't mean it's that good:) If pleo only understood only like 6 phrases.. It would be sorta weird:D!!!

mweed
02-08-2008, 10:07 PM
Six phrases is better than none! And the point is, we should be able to get pleo to recognize something. ;)

PleoPet
02-09-2008, 10:13 AM
Some comments:

Small vocabulary voice recognition toys are not new. The iCybie (circa 2001) has 8 redefinable voice commands.
They typically use a specific hardware chip to do all the work (like those from Sensory Inc).
The chips have been getting cheaper all the time, and are found in many simpler and much cheaper toys (including the newer Furbys).
http://www.sensoryinc.com/company/prod_showcase.html
----
It is possible to reprogram the Pleo head CPU to do more elaborate processing, but it will be a challenge to make a even a simple voice recognition software algorithm fit in the smaller ARM chip.
Even so the unreliable nature and small voice vocabulary will likely be disappointing for most users.
----
You could slap on a dedicated voice recognition chip and run it into the Pleo's bottom serial connector. You could also add a better microphone.
The extra hardware, unreliable nature and small voice vocabulary will likely be disappointing for most users.
-----
Finally the best experience is probably using a PC to do the voice recognition, and remotely control the Pleo with voice commands. Using a headset greatly improves the voice recognition quality. Using the computing power of a PC gives you the option of hundreds of voice commands.
This is the approach used in the RoboSapien Dance Machine software (the PC does all the work, the robot acts like a puppet under PC control)
http://www.robodance.com/

Xasher
03-03-2008, 04:36 AM
I am sure Pleo doesn't have speech recognition, but he may have sound responses. Try this experiment, when pleo is walking, say stop loud at him. I have been doing this experiment for awhile and he stops walking most of the time. Has Ugobe made Pleo so that he will respond to loud sounds while he is walking ?:cool:

Falseprophet666
03-07-2008, 12:08 PM
I remember a big guy named rad 4.0. He had a huge vocabulary. Unfortunatly, all his features were voice command stuff, but he was still pretty cool back then.

Xasher
03-08-2008, 05:09 AM
Well, roschler, once you are inside Pleo, you can write a new voice/speech recognition program, too, taking more full advantage of the under-used Pleo hardware. That way, people can converse with you, and love you for your mind, and not just your body. LOL

That is until your skin wears off and your teeth chip, then youll more likely be thrown in a closet and eventually sold at a garage sale to someone who will take you apart and cast your parts into the trash. You need to be careful what you wish for.:eek:

InmemoryofRomeo
10-23-2008, 04:52 PM
What about color/shape recognition?

Pleo is programmed to respond to the shape and/or color of his leaf isn't he? Wouldn't it be also possible to program Pleo to respond to other specific shapes and colors?

Example, walk towards a red triangle with a yellow edge.

Allosaurus
10-26-2008, 12:37 PM
If hasbro can do it fro 1/2 the price of pleo . . .

their new robotic golden retriever puppy claims recognition of six phrases

http://money.cnn.com/galleries/2008/news/0802/gallery.toy_fair/7.html

Oh, I have that toy! It works quite well, actually. :D