Operating system: Windows 10
Graphics card: Intel inside
Rendering audio from the sequencer does not render keyframes.
So, putting audio clips in the sequencer, it is possible to animate the volume (and possibly other properties). This plays back and renders to video correctly, but rendering audio causes the first value of each animated property to be used for all of it and keyframes (animation for the rest) are entirely discarded.
Just connect the included audio file to "speech.wav" audio datablock in the blend file, play it back to hear what it's supposed to sound like, then render audio from the render menu.