My fellow forum-goers;
Does Sam & Max Episode 302: The Tomb of Sammun Mak contain any PS3-Exclusive bonuses? And if so, what kind? Posthaste responses much obliged.
-Kane
Also, I hope not, because that means there will be one more thing for me to gripe about, let alone not being able to PLAY said game!
I think he's saying he can't read and uses an automatic program. That would explain why "doesn't" became "Dustin" and "didn't" became "Diderot". I assumed he was using a spell-checker due to dyslexia.
When you complete the first episode on Playstation, you get NutriSpecs in your second playthrough, which are one of Max's power that tells you what stuff is made of.
I think he's saying he can't read and uses an automatic program. That would explain why "doesn't" became "Dustin" and "didn't" became "Diderot". I assumed he was using a spell-checker due to dyslexia.
Well, now I feel like a jerk for my first post in this thread T_T...
Unfortunately speech recognition is still a bit of an unsolved problem in computer science. There isn't yet any speech recognition software that works very well. It can work OK if a very small dictionary is used--i.e. a device that recognizes only a certain number of voice commands. One example is a cell phone that allows you to call someone on your contact list by speaking their name into the microphone.
Speech recognition software that has to understand all the words in a language will invariably be inaccurate... and accents, dialects, minor speech impediments or even different speaking voices will confuse them.
Human brains will recognize and understand speech accurately and effortlessly because they are specifically wired for pattern recognition. Computers on the other hand are terrible at this. We're used to thinking of computers as very smart, as they can perform calculations billions of times more quickly than any human--but when it comes to pattern recognition they seem surprisingly stupid.
I think it's an interesting example of the distinction between human intelligence and machine intelligence. It's also kind of disappointing; we live in 2010, dammit! You'd think they would have solved problems like this years ago. Science fiction has told us that we should be visiting other planets, living on the moon and speaking to sentient computers by now.
My apologies for going off topic--I didn't mean to.
...or did I?
The only thing being voice recognition has actually been pretty awesome for years.
On anything since windows xp you can easily dictate and have it type with almost no errors because when you set it up you spend about 5 mins training it to your voice and then if it makes errors that you have to correct it further trains itself.
On anything since windows xp you can easily dictate and have it type with almost no errors because when you set it up you spend about 5 mins training it to your voice and then if it makes errors that you have to correct it further trains itself.
You're thinking of dictation software, which is trained to understand a specific speaker. I think speech recognition generally refers to a system that can be used by an arbitrary user, but I guess the term can also be used for both types.
Dictation software can work OK, but only because you can train it, as you mentioned. Training takes time though--especially as names of places, companies and people often don't exist in the dictionary. That's okay if you're just playing around, but it can get time consuming and irritating if you actually want to get some work done. And again, it will only work for one person!
In general, dictation and voice recognition software takes a lot of processing power and memory. The software is also quite challenging to develop, so I presume commercial solutions are expensive. A lot of current solutions are failures to varying degrees, e.g. automatic call center robots that don't understand your voice commands (perhaps due to how a phone line distorts sound), YouTube's automatically generated captions (mostly useless from what I've seen), supposedly automatic voice message to text translation that needs to fall back on human transcription etc.
The point is that while humans come with a very accurate built-in speech recognition system, computer software can only work OK in ideal circumstances... no accents, no weird dialects, no stuttering, lisping or other speech impediments, no background noise and so on. And even in ideal conditions there are errors. NIST has a nice metastudy thing, which shows error rates of current and past systems. It shows over 10% error rates even in the most recent systems. Numbers I've seen quoted elsewhere are in the range of 10-25%.
Ok, basically what Jake said is that the NutriSpecs being only on PS3 (for now) was a one episode in a series type thing, so for the rest of the season we can expect simultaneous psychic toys on both PC and PS3.
Ok, basically what Jake said is that the NutriSpecs being only on PS3 (for now) was a one episode in a series type thing, so for the rest of the season we can expect simultaneous psychic toys on both PC and PS3.
Actually, I think he meant The Penal Zone power was only for The Penal Zone; no other Episode. Which is sad because extra powers are fun, especially when it has nothing to do with the full plot.
Comments
Uhhhhh... no kidding...... Maybe you meant:
Also, I hope not, because that means there will be one more thing for me to gripe about, let alone not being able to PLAY said game!
Seriously, that was a stupid "question".
Are you actually foriegn, or do you just write that way?
Well, ok...
Wait. There was a PS3 exclusive?
When you complete the first episode on Playstation, you get NutriSpecs in your second playthrough, which are one of Max's power that tells you what stuff is made of.
Well, now I feel like a jerk for my first post in this thread T_T...
Speech recognition software that has to understand all the words in a language will invariably be inaccurate... and accents, dialects, minor speech impediments or even different speaking voices will confuse them.
Human brains will recognize and understand speech accurately and effortlessly because they are specifically wired for pattern recognition. Computers on the other hand are terrible at this. We're used to thinking of computers as very smart, as they can perform calculations billions of times more quickly than any human--but when it comes to pattern recognition they seem surprisingly stupid.
I think it's an interesting example of the distinction between human intelligence and machine intelligence. It's also kind of disappointing; we live in 2010, dammit! You'd think they would have solved problems like this years ago. Science fiction has told us that we should be visiting other planets, living on the moon and speaking to sentient computers by now.
My apologies for going off topic--I didn't mean to.
...or did I?
Regardless, I agree with you. Where are my totally voice-activated things? Why can't I dictate a whole novel yet? Mmh?
On anything since windows xp you can easily dictate and have it type with almost no errors because when you set it up you spend about 5 mins training it to your voice and then if it makes errors that you have to correct it further trains itself.
You're thinking of dictation software, which is trained to understand a specific speaker. I think speech recognition generally refers to a system that can be used by an arbitrary user, but I guess the term can also be used for both types.
Dictation software can work OK, but only because you can train it, as you mentioned. Training takes time though--especially as names of places, companies and people often don't exist in the dictionary. That's okay if you're just playing around, but it can get time consuming and irritating if you actually want to get some work done. And again, it will only work for one person!
In general, dictation and voice recognition software takes a lot of processing power and memory. The software is also quite challenging to develop, so I presume commercial solutions are expensive. A lot of current solutions are failures to varying degrees, e.g. automatic call center robots that don't understand your voice commands (perhaps due to how a phone line distorts sound), YouTube's automatically generated captions (mostly useless from what I've seen), supposedly automatic voice message to text translation that needs to fall back on human transcription etc.
The point is that while humans come with a very accurate built-in speech recognition system, computer software can only work OK in ideal circumstances... no accents, no weird dialects, no stuttering, lisping or other speech impediments, no background noise and so on. And even in ideal conditions there are errors. NIST has a nice metastudy thing, which shows error rates of current and past systems. It shows over 10% error rates even in the most recent systems. Numbers I've seen quoted elsewhere are in the range of 10-25%.
I just finished the 2nd Episode of the PS3 version of the game last night...
And no, I didn't unlock anything this time. So, no.
I think the NutriSpecs unlock was just a one time deal. There probably won't be anymore.
Actually, I think he meant The Penal Zone power was only for The Penal Zone; no other Episode. Which is sad because extra powers are fun, especially when it has nothing to do with the full plot.
But I hope I'm wrong, and you're right.