[vee_ess]

WELL...It is going to be a lot harder than you think it is going to be. You are not going to be able to use any chip that is out for commercial use. Maybe if you find a gov't. agency that has a chip for that, you could try stealing it and moddin it.

But anyways, the reason it is going to be harder is because you have to account for variances in sound and distinguish between various characteristics. You would have to create and design a circuit board that would calculate these, characteristics, such as voice inflection, pitch range, etc. It takes millions of lines of code for this and, frankly, it will take that much more on a circuit board.

To get around that, you are going to have to allow the device to access the hard drive or a dedicated drive (specifically for the device in question) and have it interpret and process the data.

This leads to a controller card. You can mod a controller card to do low level processing and have the data part taken care of. Now, you are goint to have to mod a chipset to use only what you need and process the data on the card which you have provided power to, the controller card (for example...a SCSI card). At that point in the design, you can reduce the processing to just what you need and fit it on the card. That brings me to another point...The controller card doesn't need to have any RAID functions, and you can get rid of any segments that process the BUS type stuff. This leads me to another train of thought...You can use RAM instead. You won't have to get rid of anything, except you will need constant power, which you will already have a need for to listen to the incoming sounds and processing them. So, basically, the RAM makes things a lot easier.

Now, for the calculations on voice characteristics...You will need to do reasearch...You will need to record, several (I can't express enough how many times), times, your voice saying "start" and "restart" several times. What you can do is just have it wait for "start" and send a signal to both jumpers, so that if the computer is not powered up, the power will be the one to do anything, and otherwise, it will reboot since the power one won't do anything because at that point it is software controlled. Back to the research and recording part - you are going to have to analyze and find all similar aspects. If you do that it makes this all much easier because then you can create circuits that, running simultaneously, analyze strength, pitch, and other characteristics. That doesn't need too much work, because you can get those things already, like at RadioShack. Then all you need to do is seperate the different segments and have it pass through one at a time.

Hope I didn't scare you at all...Just wanted to build a platform off which to start. If you need any help, please let me know because I would be very excited to work on this.

If this does work, patent it right away...become an instant millionaire!