hi,
I had try to use LM358 and TDA2822 to finish a audio circuit that shown as below.

Now, here is a trouble that a lot of noise when I using MIC input audio signal, but it looks normal work when I using my cellphone input audio signal.
there are two videos show the two condition.
Audio signal from MIC
Audio signal from cellphone
and datasheet
Datasheet of MIC

Help me to solve the trouble of noise, please, thank you very much.

 

Any noise on the 5V supply is going to go into the amplifier input when using the microphone. The phone output will be low impedance and will effectively short out that noise whne the input comes from the phone.
You will need to filter the 5V supply.
Is that supply a switch mode version? If so you might try the circuit using a battery supply. If the sound is good with a battery supply then that is definitely the cause.

 

Nobody makes a TPA2822, maybe it is a TDA2822?

Your R1 (2.2k) is too low and reduces the mic output. Use 5.6k with your 5V supply.

It is obvious that the microphone can hear sounds from the nearby speaker and the sound goes around and around called acoustical feedback squealing. Your voltage gain is 5.6 x 5.6 x 100= 3136 times! My sound level Indicator project (like a VU meter) shows a whisper in the next room with a gain of 1820 times.

Also, your R1 feeds any sounds on the 5V supply to the input of the circuit and if it is a 5V phone "charger" then it might produce the high audio frequency you hear.

The lousy old LM358 is NEVER used for audio because it is very noisy (hiss), it produces crossover distortion and has a low slew rate so it has trouble above a few kHz. Use a single low noise audio opamp instead like an OPA134.

 

The LM358 and the mic element are not biased correctly plus the gain is too high.
The schematic below shows the modifications that are required for best performance.
Remember the microphone element is polarized and must be installed correctly.

 

The LM358 and the mic element are not biased correctly plus the gain is too high.
The schematic below shows the modifications that are required for best performance.
Remember the microphone element is polarized and must be installed correctly.

The values are too high for your 4.7k resistors feeding power to the electret mic. An electret mic draws a maximum current of 0.5mA which would reduce the voltage to 5V - (0.5mA) x 9.4k)= 0.3V but an electret mic needs at least 2.0V.
I did not notice before that the original opamps were biased at 0V causing them to be rectifiers.

 

The values are too high for your 4.7k resistors feeding power to the electret mic.

I would suggest two 2.2k resistors.

 

The values are too high for your 4.7k resistors feeding power to the electret mic.

The values work fine on the breadboard they drop the voltage across the element at 2.35 volts.
I think the mic the TS is using is similar. If needed they can be changed.

 

I have used those same ICs in similar audio circuits with no noise problems. If it really is a problem, the LM358 can be replaced directly with 4558 which is low noise, with no crossover distortion.
I don't see a volume control in the circuit. I think it would more practical with a fairly high gain pre-amp, as in the original circuit but with the 10Kohm resistor on the input to the main amplifier replaced with a 10K volume control. Then the gain could be adjusted to avoid acoustic feedback from the speaker to the mic.

 

Is it noise or just overload distortion? And is the connecting cable shielded? The earphone connection from a cell phone is not at all intended to drive an amplifier. So if it works without problems when using a microphone then the problem is with the connection to the cell phone.
Certainly that amplifier is not "HiFi" but it should work if it is connected correctly.
So please let us know how it is connected. That is where the problem is.

 

The video shows the mic and speaker very close together and a high pitched oscillation (acoustical feedback) is played.
The original schematic shows the opamps biased wrong.

 

OK, I got it backwards, not realizing that post #8 was not from the TS. It is indeed astounding that so many do not understand about acoustic feedback oscillation. I am not sure how I got that reversed.
Of course, why would anybody ever have the speaker nxt to the microphone?? That makes no sense at all.. And how is it possible to confuse oscillation with noise?

 

We can learn new tricks of the trade. I think your ears have to be trained to answer why ? That needs a demonstration.
So you need to compare what sounds good to your ears. I can only guess how that processing works.
My guess is that parts of the speaker and encloure bring out effects so that secondary sound is in the background for depth.

 

Thank you very much! You pointed out key points and gave me many useful suggestions. ex. I ignored the gain of TDA2822.
According to discussion and suggestions, I had modified circuit as below for fix below items
1. UseVR1 and VR2 to adjust and reduce the total gain.
2. Adding R6 to fix the wrong of OPA biased.


The test result shown as below video
Result

It is useful for now, but is there any part that needs to be modified?

 

I suggest connecting pins 6 and 7 of U1 together so that the unused amplifier is in a stable state. Other than that it will work.

 

The sounds in the video are only high pitched squeaking. EDIT: The tiny speaker cannot produce low frequencies.
Crossover distortion produces notches in sounds like on this sinewave:

 

It is useful for now, but is there any part that needs to be modified?

See parts in the shaded boxes.
C5 and C6 to increase the low frequency response. Add 2.2K resistor to decouple mic element. Use other half of LM358 as a buffer amp.

 

It's called feedback!