but back to the original topic, how bout testing not the waveform data but the duration of the sound. though "on", "off", and "dim" last approx the same time but maybe we can work with synonym that last significantly differently.
though this seems like a crook method and will have a LOT of false triggering but hey who would want to buy sophisticated DSP chips for just three commands.
thanks!!
I really appreciate it. Anyways suppose i wnat it to recog nise my voice what parts would i need to use. And if i want it to recognise more than one voice, what do i need to do. Take ur time replyin to this question. I dont need it immediately as this is of my own interest.
well if u want that then and still be left with
a few bucks for ur next meal then i would sincerely suggest using
and old (say pentium II based) computer without its monitor,keyboard, cd-drive, floppy, maybe even harddisk etc.
this "circuit" can be built in a few hundered bucks or maybe even lesser.
and an indian (like me :twisted: ) would happily write the software to make it do what u want in a few thousand dollars. :lol:
you are so funny shima!!!
if you need to recognize a voice....
maybe only a few words, but it is so complicated, but still you cant do this with transistors(or maybe if you use about 100000?)
ow common. software rules !!!!!!!!
its aint that hard. just needs a bit of study of fourier and laplace transforms etc. i hated that topic a bit in the college and thats what justifies the Thousands of dollars i mentioned 8)
Philips manufactures Voice Recorder/Command Interpreter IC. You can program it for any words you want and whenever a voice command is spoken, the desired output will go high.