There are several methods.

adelay, amix

Use the adelay and amix filters:

ffmpeg -i audio1.mp3 -i audio2.mp3 -filter_complex "[0]adelay=4s:all=1[0a];[1]adelay=30s:all=1[1a];[0a][1a]amix=inputs=2[a]" -map "[a]" output.mp3

Note that the amix filter will reduce volume of the output to prevent clipping. Followup with dynaudnorm or volume filters if desired.

adelay, concat filter

Or adelay and concat filters. This assumes audio1.mp4 is 10 seconds long, and both inputs have the same sample rate and channel layout:

ffmpeg -i audio1.mp3 -i audio2.mp3 -filter_complex "[0]adelay=4s:all=1[0a];[1]adelay=16s:all=1[1a];[0a][1a]concat=n=2:v=0:a=1[a]" -map "[a]" output.mp3

anullsrc, concat demuxer

Or generate silent files as spacers with the anullsrc filter:

ffmpeg -f lavfi -i anullsrc=channel_layout=stereo:sample_rate=44100 -t 4 4.mp3
ffmpeg -f lavfi -i anullsrc=channel_layout=stereo:sample_rate=44100 -t 16 16.mp3

Create input.txt:

file '4.mp3'
file 'audio1.mp3'
file '16.mp3'
file 'audio2.mp3'

Then use the concat demuxer:

ffmpeg -f concat -i input.txt -c copy output.mp3