Recording. I will be rummaging through Callas’ extensive recorded legacy below, in search of evidence for my technical assessment. But to introduce the topic: While the difference between studio and live recordings is detectable with most singers, it’s especially marked in Callas’ case, and I think we can fairly say that as her career progresses, her live performances tend to more and more resemble studio recordings, rather than vice-versa. Beginning with a belief that “Records . . . are hard for me because, you know, I have the big voice” (again from the Mayer article), she comments on more than one occasion on her awareness, on the one hand, that the mike is picking up tiny artifacts that go by unremarked in the opera house, and on the other that it makes nuances of inflection and phrasing more desirable, and that these can in turn improve her opera house performance. Whether or not this awareness served her well in the long run is one of the things we’ll be pondering.

Vibrato rates. Will Crutchfield has been the point man in research on this subject, and has kindly shared much of it with me—not merely the statistics themselves, but sight/sound demonstrations of the measurements, as found through his own spectographic analysis of recordings. The basic finding: vibrato rates in Western singing voices have dropped markedly since 1900. From an average range of 6.5 to 7.5 (not infrequently higher), they have fallen to one of 5.0 to 6.0 (not infrequently lower). The evidence for this is massive, supported by hundreds if not thousands of samplings. This subsidence cuts across all voice types, all nationalities, and all vocal methods, and while it is observable from quite early in recording history, it has accelerated in the post-WW2 era, with a turbocharged plunge in the 2000s. Full elaboration of this phenomenon will have to await the completion and publication of Will’s study. But to give some context for the canary-in-the-coal-mine example of Callas, he’s laid out some 150 soprano samplings, not including those of a “pure” high coloratura type, who would not be sensibly compared to Callas. Some of the singers are given multiple samplings. They are presented in five groups, each of which represents roughly two decades of recorded evidence. To compare like with like, he has selected a single note, the high C in Aïda’s “Oh patria mia.” Summarized findings:

Group 1 (1900-1920—Boninsegna, Burzio, Destinn, Gadski, Jeritza, 20 more ): lowest 5.9, highest 8.3, preponderantly between mid-6’s and low 7’s. Group 2: (’20s and ’30s—Ponselle and Rethberg, Caniglia, Cigna, Muzio, 12 more): lowest 6.0, highest 8.4, preponderantly high 6’s to mid-7’s—actually a nudge upward among the voices cited. Group 3 (late ’40s through ’60s—Cerquetti, Gencer, Milanov, Tebaldi, 11 more): lowest 5.1, highest 7.5, preponderantly high 5’s and lower 6’s. Group 4 (later ’50s through ’70s—Caballe, Dimitrova, Freni, Nilsson, Rysanek, L. Price, 18 more): lowest 4.3, highest 7.1, preponderantly higher 5’s and mid-6’s. Group 5 (mostly 2000’s, though a couple earlier—Eaglen, Millo, Netrebko, Radvanovsky, 13 more): lowest 4.1, highest 6.8, nearly all in the 5’s, with the 6.8 in lonely splendor. In individual voices, the rates gradually decline with aging, and the correlations between age and vibrato rates are again attested to by many samplings. The numbers don’t lie, and they aren’t cherry-picked.