Sample rate conversion has up to 2 stages: multiplexing and dividing (downsample). Here we watch what changed spectrum audio for dividing frequency.
On top of the picture we see spectrum of source audio signal with sampling rate Fd. Spectrum on probation contains 2 areas (blue and light blue). We see what light blue placed above Fd/4.
We will decrease (divide 2 times) sample rate source audio and watch to metamorphoses of its spectrum. New sample rate is Fd new = Fd / 2.
How divided sample frequency (downsampling)? We remove every second sample of source signal.
Output spectrum we see on bottom of the picture. Light blue area of source spectrum mirrored on spectrum of audio with divided sampling rate. It is non-linear distortions arose in useful spectrum band after dividing sampling rate.
Possible avoid it?
Yes. Before dividing need filter (suppress) source spectrum above Fd new / 2 = Fd / 4.
About oversampling read here...