WPM Counts for YouTube Videos

wpm_counts.rb

require 'words_counted'
require 'webvtt'
require 'time'
require 'youtube-dl.rb'
require 'yt'						# JSON error with youtube-dl.rb

line_buffer = []
max_buffer_size = 5
num_words = 0
audio_start = 0
audio_end = 0
minute_counts = []

# YouTube video key
key = "8KkKuTCFvzI"

options = {
  'skip-download': true,
  'write-auto-sub': true,
  'output': "captions/#{key}"
}

YoutubeDL.download "https://www.youtube.com/watch?v=#{key}", options
webvtt = WebVTT.read("captions/#{key}.en.vtt")
webvtt.cues.each do |cue|
    puts cue.inspect
  	lines = cue.text.gsub(/<\/?[^>]*>/, "").split(/\n/)
  	lines.each do |line|
      # puts line
  		if !line_buffer.include? line
  			line_buffer << line
  			current_line_count = WordsCounted.count(line).token_count
  			num_words += current_line_count
  		  current_minute = ((Time.parse(cue.start.to_s) - Time.parse("00:00:00.000")) / 60).to_i
  			minute_counts[current_minute] = minute_counts[current_minute].to_i + current_line_count
  			line_buffer.shift if line_buffer.size > max_buffer_size		
  		end
  	end

  	# Calculate duration of audio stream
  	audio_start = cue.start if audio_start == 0
  	audio_end = cue.end
end

duration = (Time.parse(audio_end.to_s) - Time.parse(audio_start.to_s)) / 60
puts "Audio stream length (min): #{duration}"
puts "Word count: #{num_words}"
puts "WPM: #{num_words / duration}"
puts Time.at(Time.parse(audio_end.to_s) - Time.parse(audio_start.to_s)).utc.strftime("%H:%M:%S")
minute_counts.each_with_index do |value, key|
  puts "#{key}, #{value}"
end