TODO:
    specify preferred segment size?
    optionally break at sentence, punctuation
    compose scripts, e.g.:
    loop, for each segment, exportable as practice mp3:
      segment (half-rate), pause=segment duration for learner to repeat,
      repeat segment (normal rate), pause again
    auto-detect silences for rough segmentation
    display tones above characters (also pinyin, bopomofo, translation)