I think the only way is to time manually with a stopwatch. And then put some markers into the text. Something like this:
<0.01>The <0.4>quick <1.0>brown <1.6>fox <2.0>jumps <2.5>over <2.8>the <3.0>lazy <3.5>dog.
Although, I don't know if speech synthesis has the same reading speed on different devices/browsers.