Resample to 22.05 kHz
Drop an audio file and export it at 22.05 kHz as a WAV. The rate carries everything up to 11.025 kHz and is exactly half of 44.1 kHz.
Halving a 44.1 kHz source
44100 / 22050 is 2, so every output sample lands on an existing input sample. The resampler removes samples rather than interpolating between them, which means no fractional arithmetic and no interpolation error.
This is the same relationship 24 kHz has with 48 kHz. Which of the two to choose follows from the source: halve 44.1 to get 22.05, halve 48 to get 24 kHz. Crossing between the families reintroduces the awkward 160 / 147 ratio.
Memory-bound audio
Game engines and sample libraries hold large numbers of short sounds resident at once. Halving the sample rate halves the memory each one occupies, and for footsteps, impacts, and dialogue the missing top octave costs far less than the space it frees.
The same reasoning drove early multimedia audio, which is why a lot of legacy assets sit at this rate.
The 11.025 kHz ceiling
Half the sample rate is the limit. Speech still sounds natural at 11.025 kHz, since its useful content is well below that. Music suffers more visibly: cymbals lose their shimmer, and bright mixes sound dull and closed in.
Content above the ceiling is filtered out before the rate drops, so none of it folds back as aliasing.
Other delivery rates
Full-quality music wants 44.1 kHz, video wants 48 kHz, and speech recognition wants 16 kHz. For a mono downmix or another output format, use the full Sample Rate Converter.