It will be trial and error on the scroll speed. Layer markers (with comments for each line of lyrics) will be extremely useful, on the audio layer, so you can visually ‘see’ at what point on the timeline the words should appear.
Your given example, with small text, gets a lot of cushioning as there are many lyrics on-screen at the same time. For this type of scrolling technique use similar font sizes.
If you choose to chop up your lyrics there’s a handy expression that will smoothly auto-scroll the text layers:
crawlspeed = 2; value-[0,crawlspeed*timeToFrames(time)]
Apply to the Position property of your text layers.
This way, all your separate text layers will animate at the same speed.