Yeah. Lose the ball.
Animate the opacity of the overlaid text elements by character, with no smoothness, square, opacity at 0. I like to keyframe the Offset. Make the first key a hold, and the rest will follow. Keyframe according to what you see on the waveform (LL shortcut), then listen and tweak the key positions in time. Saving as a preset, or checking AE’s presets, should speed it up a bit.
Given that syllables aren’t always marked by detectable changes in amplitude, it’s folly to look for an automated solution. Ya gotta slog.