text-to-speech

Many pages are now also accessible as spoken audio.

• improve text-quality used for speech:
  • remove textual fluff (footnotes, references, tables, images, etc)
  • correct phonetics for some misspoken common words.

• research UX improvements:
  • more accurate text-audio alignment
  • word-level click navigation within the page (currently only as a separate text window)
  • auto-scroll when audio plays (as an user option) with sentence and word-level highlighting

• switch audio-codec from MP3 to Opus v1.3

• research the use of next-gen, opensource TTS systems:
  • Mozilla LPCNET
  • Mozilla DeepSpeech
  • Mozilla TTS

• find a good, portable, real-time TTS systems (for a more dynamic interaction)