You could try to use the hard paper or plastic board to isolate the speaker and microphone, if they don't work then you can using the wires to connecting speaker and microphone to the citcuit and increasing their distance, you should also put them on different direction.

 

When I change the pot (+input of the IC) I can reduce the noise. and also it appears C3 is effective in this scenario.

supply is a 9v battery

 

...you should also put them on different direction.

I cannot get this part of your statements. What do you mean different direction?

 

Reduce the microphone gain. As I have said for that reduce the pull-up resistor to 3.3 kohm. I agree that it is necessary to carry the loudspeaker and microphone at the greatest possible distance. Place the loudspeaker in a box, open on one side. Please note that the simulation I gave the signal 10 mV. Also, while I installed slider resistor 0.1 (T = .1). Gain margin is 10 times. Obscuring the speaker from the microphone with a porous material (wool, foam rubber).

 

I cannot get this part of your statements. What do you mean different direction?

Where is the sound get into the microphone and where is the sound come out from the speaker, there are two different directions, in your breadboard is the right direction, but the distance is too short may causing microphone feedback sound effects, you can get any insulator to isolating the microphone and speaker, reducing the input sound from microphone could reducing the noise, but you still have to try the maximum volume without microphone feedback sound effects, so they need to isolated or enough distance.

 

There is a simple way to reduce the sound amplification. Between the bases and collectors of transistors included resistor 330K. Replace the value of the resistor to 100k.

 

Thanks. I connected it to Vcc with a 10kohm and now it's working. but there is too much noise at output. a sound like BeeZeeZeeZee... like the sound of a Moped. I used a 10uf cap for bypass(pin number 7 of LM386) but didn't work. how to remove the noise? any idea, suggestion,...

Sounds like loading effects on Vcc are getting to the MIC.

There is about a 40% chance you could get away with no decoupling on the LM386 supply, but it'll put ripple on the supply - if that gets into the MIC bias, you'll get "motorboating".

Split the 10k bias resistor into 2x 4k7 and decouple the join between the 2 to GND with at least 100uF.

 

There is an alternative method of reducing the parasitic feedback (because of the resistance of the power supply). This swapped a microphone and a resistor. Look at the picture below. Is a variant "B". This variant does not require additional electronic components.

 

I simulated generation due to the resistance of the battery.

 

Hi again guys
Today I was looking at M851 motherboard schematic and saw an interesting circuit for microphone. look(click on the image to enlarge it):

What are those resistors and capacitors network before the mic jack?
what's the purpose of them?
which one of this lines bias the capacitor microphone?
Looks like the gain of the both Op-Amp isn't so high? am I right?