You can use the “wetware” between your ears:

Listen to the audio and do a “takedown”.

;-)

Converting audio to MIDI code is not an easy task for a computer program.

Take a pie with its several different ingredients.
Now, separate it into its constituent parts.
Vegetables, gravy, flour, water, &c.
This is, in effect, what we are asking an audio to MIDI program to achieve.

Indeed there are programs that can “do something”, but …

Simple monophonic recordings with no added reverb tend to work moderately well.
But there are still mistakes which have to be corrected manually.
Go to stereo, or add some reverb, and the results begin to deteriorate markedly.
With some simple ensembles the results begin to fall off further.
Are they usable?
It varies very greatly depending upon the instrument mix.
With complex orchestral work the results are pathetic.
Percussion within the audio files makes any conversion process even more complex.

What Peter says. What is possible depends on the complexity of the audio. Finale used to advertise this functionality with simple monophonic lines but it never worked well and was dropped with 2012 or 2014 (I forget).

Many DAWs have this functionality but again, success depends on too many factors for anyone to predict accuracy. David Das has some videos on doing this in Digital Performer, for example. Piano often works ok but there’s sometimes a lot of editing to be done. The results can be exported as MusicXML 3.0 (recommended) or MIDI. DP is not free of course but it is fully functional during the evaluation period if you want to check this out.
https://youtu.be/4oFBPI5Ro74?si=nsXKCbx5VY3LPiPt

you all are right of course, but... I will try Melodyne... anybody?

I also have Melodyne Studio. The problem with it on a mono or stereo file is that it really doesn't really know one instrument from another. This is fine if, again, a piano track but disappointing on an orchestra track. Again, you can download it for evaluation and check this out.

Something to check out are the various AI tools. These can isolate tracks and you can run Melodyne or DP on those. Many are based on Spleeter—which is free but you need to do your own user interface. The inexpensive Spleeter based options can only pull 4–5 stems from a mix, however. It's not going to pull a 3-part clarinet from an orchestra. The AI engines that Peter Jackson and Giles Martin are using on old Beatles tracks are quite costly to use and I don't think they can do it either.

I expect that Melodyne 6 and a whole bunch of others will be incorporating AI to help do what you want. I have not checked iZotope's new Ozone 11 Advanced to see if their stem separation tool is still Spleeter based for 5 stems or if it has improved. I've had 11 Advanced for a month now but have been too busy to check it out.

Thanks Mike, 

I dont know Spleeter, seems promising!

I am not able to build my own interface, I am a dummy not a developer...!

Here’s a list of 22 AI based tools including Spleeter apps. Some are sites where you upload your audio for extraction. Have fun.

https://alternativeto.net/software/spleeter/

Again, I expect that we’ll see an explosion of such apps in the coming year. Who knows? Some may prove to be useful.