Stretch/shift and combine audio/video from different sources

stargate.sh

#!/bin/bash

# deps: sox, ffmpeg

audiof="$1"
videof="$2"
outputf="$3"

tmp_file="$(mktemp /tmp/stargate.XXXXXXX)"
echo "$!" > "$tmp_file"
audio_tmp="$tmp_file.orig.wav"
shifted_tmp="$tmp_file.shifted.wav"

function myTrapHandler() {
	rm "$tmp_file"*
	exit 0
}
trap myTrapHandler SIGINT;

if [ "$#" -lt 3 ]; then
	echo "Format: $0 (audio file) (video file) (output file)"
	exit 1
fi

offset_start=0
offset_end=0
video_dur="$(ffprobe -hide_banner -loglevel error -i "$videof" -select_streams v:0 -show_entries format=duration -v quiet -of csv="p=0")"
audio_dur="$(ffprobe -hide_banner -loglevel error -i "$audiof" -select_streams a:0 -show_entries format=duration -v quiet -of csv="p=0")"
factor="$(echo "$video_dur/($audio_dur+$offset_end)" | bc -l)"
factor_r="$(echo "($audio_dur+$offset_end)/$video_dur" | bc -l)"

echo ">>> Audio: $audio_dur, video: $video_dur, factor of $factor"
echo ">>> Saving audio to $audio_tmp..."

# < /dev/null &> ~/tmp/log.txt

ffmpeg -hide_banner -nostdin -n -loglevel error -i "$audiof" -vn -map 0:a:0 -f wav "$audio_tmp"

echo ">>> Shifting (to $video_dur sec.) $audio_tmp to $shifted_tmp"
#rubberband -L --no-lamination --window-long --detector-soft --centre-focus -D "$video_dur" "$audio_tmp" "$shifted_tmp"
#ffmpeg -hide_banner -loglevel error -f wav -i "$audio_tmp" -filter:a "atempo=$factor_r" -f wav "$shifted_tmp"
soxargs="tempo $factor_r"
if (( $(echo "$offset_start < 0" |bc -l) )); then
	soxargs="$soxargs trim ${offset_start#-}"
else
	soxargs="$soxargs pad $offset_start@0"
fi
sox "$audio_tmp" "$shifted_tmp" $soxargs

rm "$audio_tmp"

echo ">>> Remuxing...."
if [ -f "$outputf" ]; then
	echo "$outputf already exists!"
	rm "$shifted_tmp"
	rm "$tmp_file"
	exit 2
fi
ffmpeg -hide_banner -nostdin -n -loglevel error -i "$videof" -i "$shifted_tmp" -c:v copy -map 0:v:0 -map 1:a:0 -movflags +faststart "$outputf"
rm "$shifted_tmp"

echo ">>> Done! Output: $outputf"

rm "$tmp_file"