Text-to-Speech for error-checking

[esbena]

Eliminate the two-person process where one reads and the other checks the typed in answers for errors: employ some text-to-speech technique.

• full blown text-to-speech is likely overkill
• upload of a few audio files and selective playback should be enough?
  • letters (for the answers)
  • numbers (for the questions)
  • "error" (signal user)
  • "next" (signal user)
  • "no answers" (signal user)
• adjustable speed
• pause functionality
• start from functionality
• only say "next" every few questions?