According to EXAMPLES-section in man-page, -h chooses highest quality. In truth it is equivalent to -q2, not -q0. Please fix EXAMPLES-section.
Man-page states "-q 0 and -q 1 are slow and may not produce significantly higher quality." [than -q 2]. With my data (six-year-old singing solo) the difference between -q0 and -q2 (at fixed-rate 128bit joint-stereo 44100 kHz) is obvious even to the layman. Since a solo-voice is hardly a symphonic orchestra, it seems misleading to discourage usage of -q0 in this way.
Hope to have been of any use.