ffmpeg

$ cat list.txt 
main.mp3|0
1.mp3|40
2.mp3|500
n.mp3|35231

$ ./ffmix.sh list.txt result.mp3 -y -b:a 64k
Done.

#!/usr/bin/bash

# Parameters:
#   $1    - input list; each line in the "file|delay" format;
#           "delay" is in seconds.
#   $2    - output file; encoder is chosen by extension, 
#           e.g. "output.mp3".
#   $3... - rest arguments are passed to ffmpeg directly.
#
# Example:
#   ./ffmix.sh list.txt result.mp3 -b:a 64k -y

i=0
inputs=()
delays=()
delayed_inputs=()
args=("$@")

while IFS='|' read file delay; do
    inputs+=("-i" "${file}")
    delays+=("[${i}]adelay=$((delay*1000)):1[${i}d];")
    delayed_inputs+=("[${i}d]")
    i=$((i+1))
done < "${args[0]}"

ffmpeg -loglevel warning \
    "${inputs[@]}" \
    -filter_complex "${delays[*]} ${delayed_inputs[*]} amix=inputs=${i}" \
    "${args[@]:2}" "${args[1]}"

echo Done.