##AudioGrep AudioGrep use https://github.com/antiboredom/audiogrep
Download audio file from youtube
youtube-dl URL --write-auto-sub -f mp3
if case download didn't specify format convert to MP3
ffmpeg -i input_video.ext -acodec libmp3lame audio.mp3
Transcibe audio to txt
audiogrep --input path/to/*.mp3 --transcribe
Create supercut of audio clips of phrases containing search words
audiogrep --input audio.mp3 --search 'pattern'
Specify output mode
audiogrep --input audio.mp3 --search 'pattern' --output-mode=word
or other options, create a regex sentence with crossfade of half a milisecond
audiogrep --input audio.mp3 --search 'artificial|city|becomes' --regex --output-mode word -o 'artificial_city_becomes.mp3' -c 500
##VideoGrep
VideoGrep use https://antiboredom.github.io/videogrep/
Convert .mpv to .mp4
https://askubuntu.com/questions/396883/how-to-simply-convert-video-files-i-e-mkv-to-mp4
ffmpeg -i input.mkv -acodec ac3 -vcodec copy video.mp4
https://ffmpeg.zeranoe.com/forum/viewtopic.php?t=468
convert .vtt to srt
ffmpeg -i input.vtt video.srt
run search on video
videogrep --input video.mp4 --search 'city'
use n-gram for occurence frequency
videogrep --input video.mp4 --ngrams 3 << number of words
PATTERNs
for a more accurate transcript use audiogrep/pocketsphinx
videogrep --input video.mp4 -transcribe
then we can do POS tags with the .mp4.transcription.txt file
videogrep --input video.mp4 --search 'JJ' --use-transcript --search-type pos
videogrep --input video.mp4 --search '^VBG DT JJ NN' --use-transcript --search-type pos
videogrep --input video.mp4 --search 'but' --use-transcript --search-type word
Incase ffmpeg crash before compiling all the temp tiles together
ffmpeg -f concat -safe 0 -i <(for f in ./*.mp4; do echo "file '$PWD/$f'"; done) -c copy output.mp4
Fillers
Youtube subtiles adds a padding to start & end time to each phrase so that the endtime of the prev phase is actually greater than then start time of the next phrase to ensure more natural clipping. Due to this overlap we can't grab fillers from this subtitle file
Use AUTOSUB for more accurate timestamp
https://github.com/agermanidis/autosub
autosub -o subs.srt -F srt video.mp4 remember to rename .srt to the same name as .mp4
SILENCE PY
python silence.py --input video.mp4 --output silence.mp4