bash script to cut mp3 silence at the start and at the end. ffmpeg silencedetect is used

ffcut.sh

#!/bin/bash

if [ "$1" == "" ]; then
    echo "no input file name";
    exit 1;
else
 mp3file="$1";
fi


echo "input "  $mp3file;
echo "output  ../$mp3file"

ffmpeg -i "$mp3file" -af silencedetect=d=3:n=-45dB -f null - 2>vol.txt

while read -r line; do
    echo "$line"

if [ -z "$sEnd" ] && [[ "$line" =~ end:.([0-9]+) ]]; then #only whole seconds are taken
       sEnd=${BASH_REMATCH[1]};
       echo found end $sEnd #first silence end
fi

if [[ "$line" =~ start:.([0-9.\]+) ]]; then
    sStart=${BASH_REMATCH[1]}
    echo found start $sStart #latest silence start
fi

done < <(grep -o -E "silence_\w*:\s-?[0-9\.]*" vol.txt)

sEnd="$(printf '%d' $sEnd 2>/dev/null)"

if [ $sEnd -gt 20 ]; then
  echo silence sStart too big $sEnd
  $sEnd="" #first silence end is too far away from start, ignoring
fi

if [ $sEnd -gt 0 ] || [ -n "$sStart" ]; then

  args=(-i "$mp3file" -vn -acodec copy)
  if [[ -n "$sEnd" && $sEnd -gt 1 ]]; then args+=(-ss $sEnd); fi
  if [[ -n "$sStart" ]]; then args+=(-t $sStart); fi
  args+=("../$mp3file")

  echo "${args[@]}"
  
  ffmpeg "${args[@]}";
  echo $sEnd - $sStart
else
  echo nothing to cut
fi

please feel free to use this script at your own risk.

It takes file name as an argument and tries to detect silence and the begin and at the end using the ffmpeg -af silencedetect=d=3:n=-45dB

Then it cuts the silence (if it has been found) using the ffmpeg -vn -acodec copy command.
It does not rebuild the file so quality is not reduced.