Files for 4577 and 4596 and new host names and images

www/eps/hpr4596/hpr4596_srt2audio.bash

@@ -0,0 +1,359 @@
+#!/usr/bin/env bash
+
+debug="1"
+padding="10"
+piper_bin="/opt/bin/piper/piper/piper"    # The location of the text to speech engine
+piper_voice="/opt/bin/piper/piper/piper-voices/en/en_US/lessac/medium/en_US-lessac-medium.onnx"
+silence_mp3="silence.mp3"
+source_volume="0"
+srt_volume="0"
+if [ "${debug}" -ne "1" ]
+then
+  ffmpeg_debug="-hide_banner -loglevel error"
+else
+  ffmpeg_debug=""
+fi
+
+#################################################
+function echo_error() {
+  echo -e "ERROR: $@" 1>&2
+  display_help_and_exit
+  exit 1
+}
+
+#################################################
+function echo_debug() {
+  if [ "${debug}" == "2"  ]
+  then
+    echo -e "INFO: $@" 1>&2
+  fi
+}
+
+#################################################
+function echo_status() {
+  if [ "${debug}" -le "1"  ]
+  then
+    echo -ne "\r$@" 1>&2
+  else
+    echo -e "INFO: $@" 1>&2
+  fi
+}
+
+#################################################
+function echo_info() {
+  if [ "${debug}" != "0"  ]
+  then
+    echo "INFO: $@" 1>&2
+  fi
+}
+
+#################################################
+function display_help_and_exit() {
+  echo_debug "For more information see "
+  exit 1
+}
+
+#################################################
+function check_variable() {
+  for variable in "$@"
+  do
+    if [[ -z ${!variable+x} ]]
+    then   # indirect expansion here
+      echo_error "The variable \"${variable}\" is missing."
+    else
+      echo_debug "The variable \"${variable}\" is set to \"${!variable}\".~"
+    fi
+  done
+}
+
+#################################################
+function timestamp_to_sec() {
+  this_timestamp="${1}"
+  this_in_sec="$( echo "${this_timestamp}" | sed 's/,/./g' | awk -F ":" '{ printf ($1 * 3600) + ($2 * 60) + $3, $0}' )"
+  echo "${this_in_sec}"
+}
+
+#################################################
+function timestamp_to_millisec() {
+  printf '%.*f\n' 0 $( echo "$( timestamp_to_sec "${1}" ) * 1000" | bc --mathlib )
+}
+
+#################################################
+function update_playhead() {
+  play_head="$( mediainfo --full --Output=JSON "${this_srt_wav_file}" | jq --raw-output '.media.track | .[] | select(."@type"=="Audio") | .Duration_String3' )"
+  echo_status "Play head: ${play_head} .. ${movie_duration}"
+  play_head_seconds="$( timestamp_to_sec "${play_head}" )"
+  play_head_milliseconds="$( timestamp_to_millisec "${play_head}" )"
+}
+
+#################################################
+function get_movie_duration() {
+  movie_duration="$( mediainfo --full --Output=JSON "${this_movie_file}" | jq --raw-output '.media.track | .[] | select(."@type"=="Audio") | .Duration_String3' )"
+  movie_duration_seconds="$( timestamp_to_sec "${movie_duration}" )"
+  movie_duration_milliseconds="$( timestamp_to_millisec "${movie_duration}" )"
+}
+
+#################################################
+function check_movie() {
+  check_variable this_movie_file
+  if [ ! -s "${this_movie_file}" ]
+  then
+    echo_error "The \"this_movie_file\" file is missing."
+  fi
+  if [ "$( file --brief --mime-type "${this_movie_file}" | grep --count --perl-regexp 'video/mp4|video/x-matroska' )" -ne "1" ]
+  then
+    echo_error "The \"this_movie_file\" variable has not a valid \"video/mp4\" or \"video/x-matroska\" mime type."
+  fi
+  get_movie_duration
+  check_variable movie_duration
+}
+
+#################################################
+function check_srt() {
+  check_variable this_srt_file movie_duration
+  if [ ! -s "${this_srt_file}" ]
+  then
+    echo_debug "No srt found trying: ffmpeg -y -hide_banner -loglevel error -txt_format text -i \"${this_movie_file}\" \"${this_srt_file}\""
+    ffmpeg -y -hide_banner -loglevel error -txt_format text -i "${this_movie_file}" "${this_srt_file}"
+    check_srt this_srt_file
+  fi
+  if [ "$( file --brief --mime-type "${this_srt_file}" | grep --count 'application/x-subrip' )" -ne "1" ]
+  then
+    echo_error "The \"this_srt_file\" variable has not a valid \"application/json\" mime type."
+  fi
+  cp  "${this_srt_file}" "${this_srt_temp_file}" 1>&2
+  if [ ! -s "${this_srt_temp_file}" ]
+  then
+    echo_error "The temp file is missing \"${this_srt_temp_file}\".~"
+  fi
+  if [ "$( file "${this_srt_temp_file}" | grep --count 'CRLF' )" -eq "1" ]
+  then
+    dos2unix "${this_srt_temp_file}" 1>&2
+  fi
+  
+  python3 -c "from bs4 import BeautifulSoup; print(BeautifulSoup(open('${this_srt_temp_file}', 'r').read(), 'html.parser').get_text())" | sponge "${this_srt_temp_file}"
+  
+  if [ "$( grep -c "⋄⋄⋄ Fin ⋄⋄⋄" "${this_srt_temp_file}" )" -eq "0" ]
+  then
+    last_srt_index="$( tail "${this_srt_temp_file}" | grep --perl-regexp '^[0-9]+$' | tail -1 )"
+
+    echo "" >> "${this_srt_temp_file}"
+    echo $(( ${last_srt_index} + 1 )) >> "${this_srt_temp_file}"
+    echo "$( \date --date="${movie_duration} today - 2 seconds" +%H:%M:%S,000 ) --> $( \date --date="${movie_duration} today - 1 seconds" +%H:%M:%S,000 )" >> "${this_srt_temp_file}"
+    echo "The End." >> "${this_srt_temp_file}"
+    
+    echo "" >> "${this_srt_temp_file}"
+    echo $(( ${last_srt_index} + 2 )) >> "${this_srt_temp_file}"
+    echo "$( \date --date="${movie_duration} today - 1 seconds" +%H:%M:%S,000 ) --> $( \date --date="${movie_duration} today" +%H:%M:%S,000 )" >> "${this_srt_temp_file}"
+    echo "⋄⋄⋄ Fin ⋄⋄⋄" >> "${this_srt_temp_file}"
+    
+  fi 
+}
+
+#################################################
+function check_mp3() {
+  if [ ! -s "${1}" ]
+  then
+    echo_error "The audio file \"${1}\" is missing."
+  fi
+  if [ "$( file --brief --mime-type "${1}" | grep --count 'audio/mpeg' )" -ne "1" ]
+  then
+    echo_error "The file \"${1}\" has not a valid \"audio/mpeg\" mime type."
+  fi
+}
+
+#################################################
+function check_wav() {
+  if [ ! -s "${1}" ]
+  then
+    echo_error "The audio file \"${1}\" is missing."
+  fi
+  if [ "$( file --brief --mime-type "${1}" | grep --count 'audio/x-wav' )" -ne "1" ]
+  then
+    echo_error "The file \"${1}\" has not a valid \"audio/x-wav\" mime type."
+  fi
+}
+
+#################################################
+function fill_silence() {
+  # ffmpeg -y -hide_banner -loglevel erroroptions
+  # -f fmt (input/output) Force input or output file format.
+  # -i url (input) input file url
+  # -t duration When used as an output option (before an output url), stop writing the output after its duration reaches duration.
+  # -q:a Set the audio quality (codec-specific, VBR). This is an alias for 
+
+  subtitle_start="${1}"
+  
+  if [ ! -s "${this_srt_wav_file}" ]
+  then
+    echo_debug "Creating the audio track \"${this_srt_wav_file}\".~"
+    ffmpeg -y -hide_banner -loglevel error -f lavfi -i anullsrc=r=22050:cl=mono -t ${subtitle_start} -q:a 9 -acodec pcm_s16le "${this_srt_wav_file}" 1>&2
+  else
+    subtitle_start_millisec="$( timestamp_to_millisec "${subtitle_start}" )"
+    subtitle_start_sec="$( timestamp_to_sec "${subtitle_start}" )"
+    if [ "${play_head_milliseconds}" -lt "${subtitle_start_millisec}" ]
+    then
+      silence_to_add="$( echo "${subtitle_start_sec} ${play_head_seconds}" | awk '{ printf ($1 - $2), $0}' )"
+      echo_debug "Adding \"${silence_to_add}\" seconds to the audio track \"${this_srt_wav_file}\".~"
+      ffmpeg -y -hide_banner -loglevel error -f lavfi -i anullsrc=r=22050:cl=mono -t "${silence_to_add}" -q:a 9 -acodec pcm_s16le "${silence_wav}" 1>&2
+      if [ ${?} -ne "0" ]
+      then
+        echo_debug "ffmpeg -y -hide_banner -loglevel error -f lavfi -i anullsrc=r=22050:cl=mono -t \"${silence_to_add}\" -q:a 9 -acodec pcm_s16le \"${silence_wav}\"" 1>&2
+      fi
+      sox -V2 "${this_srt_wav_file}" "${silence_wav}" "${this_srt_wav_file%.*}_combined.wav"
+      mv "${this_srt_wav_file%.*}_combined.wav" "${this_srt_wav_file}" 1>&2
+      rm "${silence_wav}" 1>&2
+    else
+      echo_debug "There is a overlap at \"${subtitle_start}\".~"
+    fi
+  fi
+  check_wav "${this_srt_wav_file}"
+}
+
+#################################################
+function segment_to_speech() {
+
+  this_index="${1}"
+  this_in="${2}"
+  this_out="${3}"
+  this_segment_text="${4}"
+  this_in_sec="$( timestamp_to_sec "${this_in}" )"
+  this_out_sec="$( timestamp_to_sec "${this_out}" )"
+  this_in_millisec="$( timestamp_to_millisec "${this_in}" )"
+  this_out_millisec="$( timestamp_to_millisec "${this_out}" )"
+  
+  if [ "${this_out_millisec}" -lt "${this_in_millisec}" ]
+  then
+    echo_error "The timing on segment ${this_index} is incorrect, as the end \"${this_out}\" cannot be before the start \"${this_in}\".~"
+  fi
+
+  if [ "${play_head_milliseconds}" -lt "${this_in_millisec}" ]
+  then
+    echo_debug "Adding space as the playhead \"${play_head_milliseconds}\" is behind the subtitles \"${this_in_millisec}\""
+    fill_silence "${this_in}"
+    #TODO calculate difference, Pass to new function which will create that lenght of silence and add to it.
+  fi
+  
+#   this_segment_duration_sec="$( awk "BEGIN{print ${this_out_sec}-${this_in_sec}}" )"
+#   this_segment_duration_millisec="$( printf '%.*f\n' 0 $( echo "${this_segment_duration_sec} * 1000" | bc --mathlib ) )"
+  
+  echo "${this_segment_text}" | "${piper_bin}" --model "${piper_voice}" --sentence_silence 0 --output_file "${this_srt_wav_file_next_line}" > /dev/null 2>&1
+#   check_wav "${this_srt_wav_file_next_line}"
+  
+  sox -V2 "${this_srt_wav_file}" "${this_srt_wav_file_next_line}" "${this_srt_wav_file%.*}_combined.wav"
+#   check_wav "${this_srt_wav_file%.*}_combined.wav"
+  
+  mv "${this_srt_wav_file%.*}_combined.wav" "${this_srt_wav_file}" 1>&2
+  
+  update_playhead
+  
+  rm  "${this_srt_wav_file_next_line}"
+}
+
+#################################################
+function create_audio() {
+
+  first_subtitle_time_stamp="$( head "${this_srt_temp_file}" | grep --perl-regexp ' --> ' | head -1 | awk -F ' --> ' '{print $1}' | sed 's/,/./g' )"
+ 
+  check_variable first_subtitle_time_stamp
+  fill_silence "${first_subtitle_time_stamp}"
+  
+  padding_zero="$( echo $(( $( tail "${this_srt_temp_file}" | grep --perl-regexp '^[0-9]+$' | tail -1 ) + 1 )) | wc --chars )"
+  cat "${this_srt_temp_file}" | while read this_line
+  do
+    
+    if [ "$( echo "${this_line}" | grep --count --perl-regexp '^$' )" -eq "1" ]
+    then
+      continue # Ignore Blank Lines
+    fi
+    if [ "$( echo "${this_line}" | grep --count --perl-regexp '^[0-9]+$' )" -eq "1" ]
+    then
+      # Entering a new segment 
+      # If there are no variables set then this is the first segment
+      if [ ! -z "${segment_start}" ]
+      then
+        segment_index="$(printf %0${padding_zero}d $(( ${this_line} - 1 )) )"
+        echo_debug "Generating speech for segment ${segment_index}: \"${segment_text}\""
+        segment_to_speech  "${segment_index}" "${segment_start}" "${segment_end}" "${segment_text}"
+        
+      fi
+      # Now reset the variables
+      segment_start=""
+      segment_end=""
+      segment_text=""
+      update_playhead
+      continue # Index line
+    fi
+    if [ "$( echo "${this_line}" | grep --count ' --> ' )" -eq "1" ]
+    then
+      segment_start="$( echo "${this_line}" | awk -F ' --> ' '{print $1}' | sed 's/,/./g' )"
+      segment_end="$( echo "${this_line}" | awk -F ' --> ' '{print $2}' | sed 's/,/./g' )"
+      continue
+    fi
+    # Combine multi line
+    segment_text="${segment_text} ${this_line}"
+  done
+}
+
+#################################################
+# Main
+
+for this_movie_file in $@
+do
+  echo "Processing \"${this_movie_file}\""
+  check_movie this_movie_file
+  this_movie_file_voice_over="${this_movie_file%.*}_voice_over.mkv"
+  this_movie_original_mp3_file=".~${this_movie_file%.*}_original.mp3"
+  this_movie_new_mp3_file=".~${this_movie_file%.*}_mixed.mp3"
+  this_srt_file="${this_movie_file%.*}.srt"
+  this_srt_temp_file=".~${this_srt_file%.*}_temp.srt"
+  check_srt this_srt_file
+  this_srt_wav_file=".~${this_srt_file%.*}_srt.wav"
+  this_srt_wav_file_next_line="${this_srt_temp_file%.*}_next_line.wav"
+  silence_wav=".~silence.wav"
+  play_head="00:00:00.000"
+  play_head_seconds="0"
+  play_head_milliseconds="0"
+
+  if [ ! -e "${this_srt_wav_file}" ]
+  then
+    echo_info "Creating the new media"
+    create_audio
+  fi
+  
+  echo ""
+  echo_info "Extracting the original sound track from \"${this_movie_file}\" to \"${this_movie_original_mp3_file}\""
+  ffmpeg -y -hide_banner -loglevel error -txt_format text -i "${this_movie_file}" "${this_movie_original_mp3_file}"
+  
+  if [ "${source_volume}" != "0" ]
+  then
+    echo_info "Changing volume of movie \"${this_movie_file}\" to \"${this_movie_original_mp3_file}\""
+    sox -v ${source_volume} "${this_movie_original_mp3_file}" ".~${this_movie_original_mp3_file}"
+    mv -v ".~${this_movie_original_mp3_file}" "${this_movie_original_mp3_file}"
+  fi
+  
+  if [ "${srt_volume}" != "0" ]
+  then
+    echo_info "Changing volume of subtitle audio \"${this_movie_file}\" to \"${this_movie_original_mp3_file}\""
+    sox -v ${srt_volume} "${this_srt_wav_file}" ".~${this_srt_wav_file}"
+    mv -v ".~${this_srt_wav_file}" "${this_srt_wav_file}"
+  fi
+
+  echo_info "Combining \"${this_movie_original_mp3_file}\" and \"${this_srt_wav_file}\" into a new track \"${this_movie_new_mp3_file}\""
+  ffmpeg -y -hide_banner -loglevel error -txt_format text -i "${this_movie_original_mp3_file}" -i "${this_srt_wav_file}" -filter_complex "[0:a][1:a]amerge=inputs=2[a]" -map "[a]" -ac 2 "${this_movie_new_mp3_file}"
+  
+  echo_info "Adding \"${this_movie_new_mp3_file}\" to \"${this_movie_file}\" to give \"${this_movie_file_voice_over}\""
+  ffmpeg -y -hide_banner -loglevel error -txt_format text -i "${this_movie_file}" -i "${this_movie_new_mp3_file}" -map 0 -map 1:a -c:v copy "${this_movie_file_voice_over}"
+  
+  if [ "${debug}" -eq "0"  ]
+  then
+    rm "${this_movie_original_mp3_file}" 1>&2
+    rm "${this_srt_temp_file}" 1>&2
+    rm "${this_srt_wav_file}" 1>&2
+    rm "${this_movie_new_mp3_file}" 1>&2
+  fi
+
+  cp -v "${this_srt_file}" "${this_srt_file%.*}_voice_over.srt"
+
+  echo_info "Finished generating \"${this_movie_file_voice_over}\""
+done