Using multiple microphones

Hi, noob here. I am trying a project where I wire up my room with multiple mics. Using a speech recognition library, identify commands and control things like Lighting, Music, etc. I want the recognition to work anywhere in the room, so using one Electret mic won't be enough. Is there a way, I can use four or five different mics, where the speech library detects the commands from all mic sources (or whatever it can detect from)?

Sure just amplify each microphone and put it into a mixer like this:-

Using a speech recognition library,

What one would that be then? You can’t do that on an Arduino Uno.

Im sorry, im new to electronics, do you mind explaining the drawn circuit?

The triangle is an operational amplifier. It has two inputs + and - with one output. https://en.wikipedia.org/wiki/Operational_amplifier

The wiggle lines are resistors. https://en.wikipedia.org/wiki/Resistor

The circuit works through summing current to a virtual ground at the - or inverting input.

Im sorry, im new to electronics, do you mind explaining the drawn circuit?

What sort of answer did you expect?

Read here, very good explanation of the inverting amplifier and its gain. Adding additional resistors lets multiple inputs be added together. http://masteringelectronicsdesign.com/how-to-derive-the-inverting-amplifier-transfer-function/

Thank you for the replies. This is a great place to start!

From a PM - the OP asks:-

I noticed you said I couldnt use an arduino uno? How come?

Because the Arduino does not have enough memory or processing power to do any sort of speech recognition. The best you can do is to attach a shield with a better processor on it and let that do the work. However it is not very good.

How about using a Raspberry Pi for this:- https://diyhacking.com/best-voice-recognition-software-for-raspberry-pi/

It's still not good but better than a shield. Mind you have you seen Siri that is hopeless and it has cloud backup.

Good point about Siri. Voice to txt over phone seems not too bad, autocorrecting gives odd results. And that's with a mic next to one's face. Whole house commands need more than a 16 MHz 8-bit processor.

Hi, noob here. I am trying a project where I wire up my room with multiple mics.

I'm not sure how much that's going to help unless you have a very large room, or unless you are planning on speaking directly into one of the mics.

The blended sound of 4 or 5 mics scattered around the room isn't much different from one mic.

Even using one of the Electret mics? My impression, was one wasnt powerful enough. I do have a fairly small room. You think it is enough?

My impression, was one wasnt powerful enough

Powerful? Do you mean sensitive?

A friend of mine -a Cambridge Maths graduate - spent months trying to do voice recognition with a very fast PC - and decided it was very difficult, and was never happy with his results.

You've absolutely no chance with a (nice but slow) microcontroller such as an Arduino...

sorry

Allan

Multiple mics in a room to have more coverage and connected to one speech processor won't work. Unless maybe you train/store the same word/sentence as multiple commands. Even with one mic you have trouble close-up if you have trained a command at 3meters distance. I have trained words into my EasyVR shield at a distance of 2-3meters, and that works 90+% of the time. Move it to another room (different reverb times), and recognition drops to almost unusable. Leo..