It is now possible to convert a normal speech recording into a studio recording.
Everyone knows that the auditory channel of human perception of information is much smaller than the visual channel of perception, and therefore interference in the perception of audio information more significantly wastes the computing resources of a person who is trying to understand the meaning in a noisy recording of a lecture or any other audio information.
Here is an example of a solution for extracting speech from noise and recovering speech
Of course, there are solutions that allow you to record speech outside of studio quality, but they will require a small outlay for a microphone and for voice processing in the recording process.
You need to buy a hardware (CEDAR DNS 2 dialogue noise suppressor) or software (CEDAR DNS Two) CEDAR noise reduction, or buy Universal Audio interface including UAD-2 DSP Accelerator, for example UAD-2 SATELLITE THUNDERBOLT 3 for $1200.00 + their C-Vox plug-in (C-Suite C-Vox Noise & Ambience Reduction | UAD Audio Plugins | Universal Audio) for $349.00. Of course in addition you need more refinements of saturation, equalization, compression - but this is the Stone Age, you need to use actual neural network solutions, not metal coffins, which work fine, but not everyone can afford them.