Andy, when I first got the spectrum analyzer (plug-in for Sound Forge) ... I went nuts and made small .wav files of all my favorites and then stored the sonograms (I since lost all in a hard drive "crash").
In particular, I remember comparing Sarah Vaughn to Ella Fitzgerald ... Vaughn's had a much longer onset time (note held longer before applying the vibrato). Once she started though ... the extent was much greater and rate was a tad slower.
As Steve pointed out, the lower the frequency ... the wider/slower the vibrato. Comparing folks like Mel Torme to Judy Garland was really enlightening.
A whole number of waves/cycle and an even-ness between the maxima (sharp) and minima (flat) combined with variations of onset and rate were the hallmark of the greats.
Sampling poor examples (my own playing) showed irregular wave functions (ie. not smooth sine curves), deviations from the "dead center" of the note being embellished ...
And worst of all ... changes in rate that were inconsistant. Mine would ... start fast, then slow down, then speed up ... Yuk
The exciting players like Andy Iona will vary the rates .. but always in a "build-up" way. A "pure" note followed by an increase in both rate and extent .... very hard to tell in "Real-Time" (easy to see in the 3K Hz range of a sonogram) ... but flat out thrilling to hear. I love that era !!!
Here's one from that tune on my site ... its a 6 Hz ; 50 cent vibrato ...
Hey ... I'm workin' on it
------------------
<font size=1>www.horseshoemagnets.com </font>