Simultaneous audio processing from multiple speakers

Hi. First i must confess that i'm a newbie and so this question might seem stupid... ;D

But do you guys think that it is possible to create a system using the scheduler library to process sound from multiple microphone inputs.

I'm working on a project to alert my school in case of a human emergency and the idea is to send multiple sound inputs from multiple sound sensors to the arduino and to sound the alarm when a key word like "HELP" is detected using the arduino voice shield.

Would i be able to this with the scheduler library?.... if not are there any ideas or tutorials out there about how i can do this? Can you Pls point me in the right direction... i really need help.

I've never done it, but people have done speech recognition with the Arduino. [u]This[/u] was the 1st hit I got with Google.

I don't understand what scheduling has to do with this unless your emergencies are scheduled... :smiling_imp:

I have a few MAJOR concerns but the one thing you have going for you is a limited vocabulary.

My concerns are:

  • What happens if somebody says, "I need help with my homework"?

  • There could be other false positives where the Arduino "thinks" it hears "Help" (as well as misses). Humans only hear about 80% of words correctly before our brains figure it out from the context, and I'm pretty sure computer technology has not caught-up with humans yet.

  • Multiple unknown speakers. Voice recognition works best when the software is "trained" to recognize a particular speaker.

  • The Arduino has limited processing power compared to something like a PC running Dragon Naturally Speaking.

  • Speech recognition works best with consistent volume levels and a good signal-to-noise ratio. (Someone speaking clearly and directly into a microphone in a reasonably quiet environment.)

  • Do you really want the whole school relying on a home-built alarm system?

  • Do you really want the whole school relying on a home-built alarm system built by a "newbie"?

In addition let me say you have absolutely no chance of making this work.

Thanks for your response

Grumpy_Mike: In addition let me say you have absolutely no chance of making this work.

Why??

Why??

Both DVDdoug and Mike are saying the same: the only difference is that, say, Mike is more direct ;)

Regards

DVDdoug: I've never done it, but people have done speech recognition with the Arduino. [u]This[/u] was the 1st hit I got with Google.

I don't understand what scheduling has to do with this unless your emergencies are scheduled... :smiling_imp:

I have a few MAJOR concerns but the one thing you have going for you is a limited vocabulary.

My concerns are:

  • What happens if somebody says, "I need help with my homework"?

  • There could be other false positives where the Arduino "thinks" it hears "Help" (as well as misses). Humans only hear about 80% of words correctly before our brains figure it out from the context, and I'm pretty sure computer technology has not caught-up with humans yet.

  • Multiple unknown speakers. Voice recognition works best when the software is "trained" to recognize a particular speaker.

  • The Arduino has limited processing power compared to something like a PC running Dragon Naturally Speaking.

  • Speech recognition works best with consistent volume levels and a good signal-to-noise ratio. (Someone speaking clearly and directly into a microphone in a reasonably quiet environment.)

  • Do you really want the whole school relying on a home-built alarm system?

  • Do you really want the whole school relying on a home-built alarm system built by a "newbie"?

With the problem of false recognition, i had an idea of using the arduino voice shield to create a voice table of the most probable alert phrases.....

but since you guys say it wont work i think ill work on something else .... something simpler... I think i was swayed by newbie enthusiasm :roll_eyes: ;) Thanks anyway.