A starting point for pitch, formant and breath reduction: filters on Tsukuyomi-chan's voice, picked by ear.
original = her own voice; soft (the default) = a little calmer and lower; low, deep, deeper = progressively deeper.
Choosing one moves the sliders below; moving a slider afterwards overrides the preset.
Sentences with no Japanese at all (e.g. "Let's do our best!"). read = the model reads them in its English
mode, with a Japanese accent; skip = left out. Mixed sentences like 「今日はmeetingです」 are always read.
Off: pressing Speak again stops what is playing and starts the new text. On: the new text waits until the
current speech has finished.