I'm Currently doin a project using voice recognition to unlock a door. Im planing to use the PIC16F877 microcontroller.Is it possible?? Im havin difficulty to start on the microphone circuit, the filter circuit which allow the range of 300-3400hz to pass through. Which kind of filter should i use. Im also havin dificulty on the Aanalogue to Digital conversion of voice to allow the microcontroller to recognize the input. How should i build it?? Finally im using mplab to program the microcontroller for it to recognize the input voice of 80% accuracy to unlock the door? Do u hav any idea how should i wrote it??

 

There are commercially available voice recognition modules available, I'm rather dubious about trying to use a PIC to do it!.

Certainly if you are struggling with simple filtering, the project sounds rather complicted for you - all you need is a bandpass filter, or a lowpass and a high pass - there are certainly plenty of circuits out there for voice bandwidth filters, try amateur radio sites.

__________________
PIC programmer software, and PIC Tutorials at:
http://www.winpicprog.co.uk

 

thats a big job for a PIC. voice recognition needs complex algorithms. a DSP or an FPGA might be able to do this job.

 

Yikes... if it were me, I'm not so sure I'd want to use voice recognition. There could be too many flaws in it. If you're looking for a keyless/codeless solution, maybe you should go for fingerprint reading. But, that's me. 8)

__________________
When life gives you lemons... make a battery.

 

As Nigel pointed out ther are commercial voice recogintion modules at remarkable low prices to be found. Not that $50 american is cheap, but considering the small miracle it has to perform it is not that bad


It may be to much for a pic on its own, however many years ago I do recall seeing the hex code for speech recognition using the humble ZX81, surprisingly this humble machine still has a few devoted owners on the web. They might just might have an idea or two how it could be done with a microcontroller or try bouncing the idea off one of the PIC forums that abound on the web.

 

there's a company called sensory that makes a drop in voice recognition chip. Writing your own voice recognition code is really not an option - the required level of knowledge on human speach is really high.

Brent