This post is a 'by request' from some of the other members, extending our vocal acoustics analysis to one of the most famous arias in the Tenor repertoire, 'Pour Mon Ame' from Gaetano Donizetti's opera La Fille du Regiment (aka, the Daughter of the Regiment). The note which we will examine is the last high C in the aria. FYI, this aria is famous for having NINE of them :-)
Along the way, I will extend some of my comments made in an earlier blog post about 'Engineering Artifacts' that can work their way into a recording through the use of EQ applied during the recording or production process. In the case of two of the recordings, the effect is clearly visible to different amounts.
The tenors I have selected for this comparison are Juan Diego Flores and Luciano Pavarotti, both of whom have enjoyed success with this aria and the stage role. Each recording was taken from a live, staged performance before an audience.
Here is the spectragraph of Flores' two renditions. Since this note is sung unaccompanied, I was able to use the noise floor (that is, the loudness of the silence) between the first 2 harmonics to match the relative volume of the performances.
The 'blue' tracing is from youtube at
The 'white' tracing is from
If you recall from my blog post from a few weeks ago, I mentioned that one of the artifacts of EQ tweaking is a 'too regular' appearance of harmonics in the display. Normally, each harmonic has its own intensity, and we do not see regular 'curves', 'arcs' or 'flat spots' in the response. We do expectto see peaks when we compare harmonic-to-harmonic, as this is a characteristic of vocal formants and resonance tuning strategies by singers. We do not expect to see the noise floor rise appreciably, and then level off.
With that in mind, examine the white and the blue tracings.
In the neighborhood of harmonic 1 (fundamental, H1) and H2, the noise floor ( the low spots between the peaks) is almost identical, and the peaks of those two harmonics closely resemble each other in intensity, with a little advantage to the white recording. Perhaps the recording mic was a little closer, picking up the ambient noise at about the same level, but a bit more of his voice. That is to be expected when a staged performance (not a studio recording) is being used.
Now, look at how the white and blue lines diverge between H2 and H3, not just the peaks, but the troughs between them. These areas are elevated in both recordings, but the white recording more than the blue. I think this is an indicator that both of the recordings got some EQ help in this frequency range, with the white recording receiving more than the blue.
For comparison, here are the same notes from Luciano Pavarotti, both from live performances, in white at Covent Garden, and in Blue, at the Met in NYC.
At the Met:
and at Covent Garden
In these two recordings, there is very little change in the noise floor across the displayed frequency range, though it looks like there is just a bit more room noise in one. Neither one of these two recordings show any artifacts of EQ boosting at all, perhaps because they are already pretty good at vocal clarity. Since the white trace noise floor actually softens, perhaps the engineers turned him down a bit above 2500 Hz. :-)
It is interesting, too, to see the effect of performance miking on the audio. To the ear, the Covent Garden recording sounds a little less 'present', as if the mic is farther away from the stage, or off-axis from the soloist. You can readily see that the blue recording (from the Met) shows substantially more energy in the singer's formant region, which (for Pav) is harmonics four to six.
These four recordings (two of Flores, and two of Pavarotti) show that they both use the 2nd-formant tuning to the 3rd harmonic (as discussed a few posts ago) and also singer's formant, the combined resonance strategy.