Convert Audio from MP4 to Text

convert.rb

#!/usr/bin/env ruby

# Install some tools first
# brew install ffmpeg
# brew install lame
# brew install --HEAD watsonbox/cmu-sphinx/cmu-sphinxbase
# brew install --HEAD watsonbox/cmu-sphinx/cmu-pocketsphinx

# Place this file in the directory that you have files to convert.
# Make this file executable
# chmod +x convert.rb

# Let er rip!
# ./convert.rb
TYPE_OF_FILE_TO_CONVERT = "*.mp4"

def convert
  Dir.glob(TYPE_OF_FILE_TO_CONVERT) do |file|
    @file = file

    # This step can be commented out if your file is already an MP3
    convert_file_to_mp3(file, output)

    add_silence_to_audio_file(output, file_with_silence)
    remix_file_to_sanitize_headers(file_with_silence, file_with_remix)
    convert_to_wav(file_with_remix, wave_output)
    convert_audio_to_text(wave_output, text_output)
    cleanup([output, file_with_silence, file_with_remix, wave_output])
  end
end

def add_silence_to_audio_file(input, output)
  `ffmpeg -i concat:'silence.mp3|#{input}' -codec copy #{output} -y`
end

def create_silence
  `ffmpeg -f lavfi -i aevalsrc=0:0::duration=10 -ab 320k silence.mp3 -y`
end

def convert_file_to_mp3(input, output)
  `ffmpeg -i #{input} -b:a 320K -vn #{output} -y`
end

def remix_file_to_sanitize_headers(input, output)
  `lame #{input} #{output}`
end

def convert_to_wav(input, output)
  `ffmpeg -i #{input} -acodec pcm_s16le -ac 1 -ar 16000 #{output} -y`
end

def convert_audio_to_text(input, output)
  `pocketsphinx_continuous -infile #{input} -logfn /dev/null > #{output}`
end

def basename
  File.basename(@file, '.*')
end

def output
  basename + '.mp3'
end

def file_with_silence
  "with_silence_#{basename}.mp3"
end

def file_with_remix
  "remix_#{basename}.mp3"
end

def wave_output
  basename + '.wav'
end

def text_output
  basename + '.txt'
end

def cleanup(array_of_files)
  array_of_files.each do |file|
    `rm #{file}`
  end
end

create_silence
convert