Request for help | AI

[altjen]

Hi.
I was wondering creating a Artificial intelligent program. I know that all app wo do even simple tasks are AI. but I dont want a simle task app. I was wondering for much more. speech recognation, do simple tasks. and why not even speak to me back. (I know how to do this. but need to link this with speech recognation. this is the part where I got no idea how to do so)

PS: btw was thinking. is it possible for a software follow my command. search on google. find information and save that in a .text file or microsoft word file?

and after that after I request for such infomation it show me?

[Schol-R-LEA]

These really aren't things an individual programmer - no matter how skilled - can accomplish on their own. While there are existing natural language, text-to-speech, and speech-to-text engines available, most of them are commercial libraries which have been developed over years or even decades by large teams of developers. You may be able to use one of these for the interface to your program, but then writing the program which interprets these for meaning is itself a major undertaking, and I know of no commercial or open source engines that provide such services - the response engines of programs like Siri or Watson are for the most part closely guarded trade secrets.

In any case, none of them are really intelligent or self-aware, and their limits are significant. This is an open research problem that may require solving the Hard AI Problem (that is, it may only be possible with a genuinely sentient intelligence equal to or greater than that of an intelligent human, something that may or may not be possible).

[Minimalist]

There is lot of work beeing done in this field of computing. Adding to what Schol-R-LEA has posted, do a google search on:computer voice commands and also look at this article:
http://www.pcworld.com/article/2055599/control-your-pc-with-these-5-speech-recognition-programs.html

[altjen]

I dont need to control my pc. I just want him do simple tasks. as search on google. understand what I mean and reply. things like that..

[Schol-R-LEA]

You said that you wanted it to be able to perform a search and retrieve relevant data from that search. That's the difficult part. Programs don't actually understand anything, at least not any programs currently in existence; sentience would be required for that. Current SOTA programs can fake it pretty well, but as I said, that's beyond even the best individual programmers to write on their own.

[altjen]

guess I will waste lots of time working on this but fuck it. I got some ideas. anyway thanks for help. Take care

[Schol-R-LEA]

Well in that case, if you are determined to try, far be it from me to discourage you, if only because you may find something no one else has yet. I would recommend you consider the difficulty what you intend to do long and hard, be prepared to do a great deal of research into prior art before you begin, and be ready to write it off as a learning experience when all is said and done. Learning is rarely a bad thing, after all.

[Mr.M]

Well as you said you want your computer to understand you and be able to reply back "I hope you understand that very well" because the computer or a software doesn't know how to reply and when and with what to reply with, so this is a very time consuming project and I personally think you can achieve at least not perfect but close depending on how badly you willing to achieve it and time you spend on it. First of all I will have to advice you to first write a program flow and also as you said you want yours to do the "Simple" tasks, you will have to write the things you will be saying to it, and also write the response/replies you expect.

After that you will need to make your Voice Recognition (VR) and if you will use codecs you will need a very good codec so that your program will be able to at least getting as close as what you have spoke, because programs usually fail to correctly detect what you said.

After that you will have to now plan how you want to perform the response like suppose you said "Computer google search facebook" then you will need to make it detect that the recording has these words 'google search', 'facebook' where 'google search' will be telling a program that it will be searching online and the search engine it will use is google and 'facebook' is the parameter or value it will be searching for. But I think you can achieve much better if you first make your program recognize your voice and also be able to write what you have said as a start to make sure its does get what you said.

This is a very huge project and that is why I strongly suggest to write down everything before you jump to coding it because there will be a lot of things you will decide to change within you idea.

[altjen]

Mr.M thank you for your idea of desing. I WANT IT REALLY BADLY. Im 18 years old and study in a school that we dont learn programming. Im in final year so usually Im learning for final exams every day since 6-7 pm till 12 pm (usually) since 12pm till 2-3 am I design softwares. Ideas and projects.
With this project Im thikinking something really huge. Something that if will be complied perfectly as I'm thinking it will change the world.

But your idea of helping for searching is really amazing and different from what I thought. I like that. Thank you very much.

as I said before I dont have much time now to search for VR so can you please help me with any link to follow or anything else?

and yea you was right. I need to write everything down before I go.

thanks again for everything :)

[JamesCherrill]

Your enthusiasm and commitment are truely superb. I'm sure that you have the right attitude to achieve great things.
But in this case I agree that this total project is unrealistic, and could just demotivate you. Apple shipped the first versions of their voice command software in MacOS around 1996. Ten years and billions of dollars later it kind-of working OK with Siri, but still not that great. There's only so much one man can do.

Why not break it down into components, and just work on one of them for now. Maybe try the part that takes some input and parses it to understand what it means in terms of doing web searches or whatever. You can do that with simple typed input at first, and front-end it with voice recognition later, as a separate sub-project. (Or maybe collaborate with others to work on sub-projects in parallel.)

Yes, you want to do something amazing that will change the world. And I hope you succeed. All I'm saying is take it one step at a time and get some successful pieces completed one at a time so you can show progress and enjoy the rush you get when something works.

[altjen]

That is true James. But in 1450 AC people imagine and loved to fly. to go there where birds go. In 1945 AC super power countries as USA Germany and Russia tried to capture the orbit. They all failed at that time. now we have air planes to send us where birds cant go. and we are sending satelites in other universe. landing in other planets etc. all what I am trying to say is that they did not tried as well. Some will say they had no technology they needed. Da Vinci did not. but he made the first successful flying machine.

I dont want to look as a naive person. I am realistic person. I know this what I am doing is WAY BEYOND my skills, for now of course.

probably I said wrong that "it will change the world" but it will give huge impact in.

The software Im trying to do is something like. hear me what I say. understand it by the way as Mr.M said. with all data he have on database he find the best and short way to do something.
I know it is not too much AI. but for now I would be really glad and proud for my self if will do that.

[JamesCherrill]

Please don't let me put you off. The world we know was made by people who didn't listen to the no-sayers. Go for it! I wish you all the luck in the world.
J

[altjen]

thanks for the wishes James, but I am not that type of person who believes in luck. I dont except getting things from luck or god. God create us and Im very thankful to him. but you cant get a flower without throwing seeds in the dirt ;)

[Mr.M]

Just the links you can check out. Please ensure you read first its post and see which parts are more close to what you want to achieve. The first link has a voice recognition which you can try changing to achieve at least picking up what user has said. I think you can have a collection of commands as texts or voice clips and for each record received the application will loop through the collection on commands and see if it has any match if it does you can then either diagnose the task like opening a default browser and submit the command to it.

This project is kinder motivating me, I will also try working out something but one task at a time because this is a complex project. I will also encourage you to use codecs to compress the audio(recordings) but use a good one which will produce clean recordings.

http://www.codeproject.com/Articles/17728/Speech-Recognition-And-Synthesis-Managed-APIs-In-W

http://www.codeproject.com/Articles/579471/How-to-Write-Your-Own-Siri-Application-Mobile-Assi

This is as close as possible to what I just said of having a collection of commands so I will suggest you start here.

http://www.codeproject.com/Questions/413665/Speech-recognition-in-vb-net-without-that-annoying

Excellent sample here.

http://www.codeproject.com/Articles/483347/Speech-recognition-speech-to-text-text-to-speech-a

You obviously will need to download a C# to VB.NET converter or use an online converter to convert it.

Hope this helps. I will also try working on this when I get sometime because I see a possibility of achieving this.

[Mr.M]

This will give you motivation against those said it not practical for an individual to make this. check out this "Amy" example video and see. https://www.youtube.com/watch?v=M1YfjY8jcLg this is just to keep you inpired and motivated that you too can achieve this.