scripts/record-audio.sh
2025-04-13 23:26:39 -07:00

94 lines
3.0 KiB
Bash

#!/bin/sh
# Version 1.2
# shoutout to https://gist.github.com/Cephian/f849e326e3522be9a4386b60b85f2f23 for the original script,
# https://github.com/xythh/ added the ankiConnect functionality
# toggle record computer audio (run once to start, run again to stop)
# dependencies: ffmpeg, pulseaudio, curl
# where recording gets saved, gets deleted after being imported to anki
DIRECTORY="$HOME/.cache/"
FORMAT="mp3" # ogg or mp3
# cut file since it glitches a bit at the end sometimes
CUT_DURATION="0.1"
#port used by ankiconnect
ankiConnectPort="8765"
# gets the newest created card, so make sure to create the card first with yomichan
newestNoteId=$(curl -s localhost:$ankiConnectPort -X POST -d '{"action": "findNotes", "version": 6, "params": { "query": "is:new"}}' | jq '.result[-1]')
#Audio field name
audioFieldName="SentenceAudio"
#if there is no newest note, you either have a complete empty anki or ankiconnect isn't running
if [ "$newestNoteId" = "" ]; then
notify-send "anki connect not found"
exit 1
fi
if pgrep -f "parec"; then
pkill -f "parec"
else
time=$(date +%s)
name="$DIRECTORY/$time"
wav_file="$name.wav"
out_file="$name.$FORMAT"
if ! [ -d "$DIRECTORY" ]; then
mkdir "$DIRECTORY"
fi
notify-send -t 1000 "Audio recording started"
#timeout 1m arecord -t wav -f cd "$wav_file"
# just grabs last running source... may not always work if your pulseaudio setup is complicated
if ! timeout 1m parec -d"$(pactl list sinks | grep -B1 'State: RUNNING' | sed -nE 's/Sink #(.*)/\1/p' | tail -n 1)" --file-format=wav "$wav_file"; then
notify-send "Error recording " "most likely no audio playing"
rm "$wav_file"
exit 1
fi
input_duration=$(ffprobe -v error -select_streams a:0 -show_entries stream=duration -of default=noprint_wrappers=1:nokey=1 "$wav_file")
output_duration=$(echo "$input_duration"-"$CUT_DURATION" | bc)
# encode file and delete OG
if [ $FORMAT = "ogg" ]; then
ffmpeg -i "$wav_file" -vn -codec:a libvorbis -b:a 64k -t "$output_duration" "$out_file"
elif [ $FORMAT = "mp3" ]; then
ffmpeg -i "$wav_file" -vn -codec:a libmp3lame -qscale:a 1 -t "$output_duration" "$out_file"
else
notify-send "Record Error" "Unknown format $FORMAT"
fi
rm "$wav_file"
# Update newest note with recorded audio
curl -s localhost:$ankiConnectPort -X POST -d '{
"action": "updateNoteFields",
"version": 6,
"params": {
"note": {
"id": '"$newestNoteId"',
"fields": {
"'$audioFieldName'": ""
},
"audio": [{
"path": "'"$out_file"'",
"filename": "'"$time"'.'$FORMAT'",
"fields": [
"'$audioFieldName'"
]
}]
}
}
}'
# opens changed note, comment if you don't want it.
curl -s localhost:$ankiConnectPort -X POST -d '{
"action": "guiBrowse",
"version": 6,
"params": {
"query": "nid:'"$newestNoteId"'"
}
}'
notify-send -t 1000 "Audio recording copied"
rm "$out_file"
fi