Studio Recordings vs. Live Recordings

Recordings of complete operas fall into two main categories: studio recordings and "live" ones. In recent years the cost of making studio recordings has risen dramatically, while the executives in charge of record companies have concentrated on turning a profit in a short time, so studio recordings have become very rare. Many famous recordings of the 50's and 60's have been in continuous release ever since they were first published, amortized their costs long ago, and have been profitably sliced and diced into dozens of compilation discs. But the industry—or what remains of it—has now adopted the business model of popular music, and if a recording doesn't sell a lot of copies very fast, it isn't considered viable. Complete operas do not sell enough copies quickly enough to fit this business model. By default, most recordings of complete operas are now live, most often filmed in performance in the opera house and sold on DVD. Films of operas are another matter. Lip-synching singers we'll leave for another time.

While it's sad that studio recordings have become so rare, the comparative abundance of live performances has created a vigorous new market for "real" performances. Whereas studio sets usually had an active intention to create an "ideal" or even "definitive" version of a given work, filmed opera performances are often "records" in another sense: records of specific productions. The contemporary contributions of set and costume designers are rarely directly derived from the composer's written instructions. And when the stage director gets busy, it's much more common for the composer's stated desires to be ignored, if not actively contradicted, then for them to be respected. What results is a film of a particular interpretation of a work. I would argue that that's a very good thing, although it's obvious that not every filmed production is a success.

Studio recordings began in the early years of the 20th century, and were limited in many ways. The singers voices were the main objects of attention, and because of the primitive conditions under which these sets were made, very little actual orchestral sound made its way onto the shellac sides eventually offered for sale. Because only a short amount of music could be recorded, sometimes as much as 4 minutes, but often as little as 2 or 3, musical continuity in playback was impossible. Apart from the technical difficulty of producing sides that accurately matched up, the listener had to stop the record after each side, carefully remove that disc (or cylinder) and replace it in its protective packaging, take out the next record and carefully place it on the turntable and start the music again. All of this took time, and the sense of a complete performance was difficult to obtain, at best.

All this changed by the early 50's, and with the flood of complete sets that poured forth came a new attitude. Producers now wanted to capture not just a performance, but something that summed up the composer's intentions. We weren't buying just another performance of "Carmen", we were being offered "Carmen" as an object that came as close as possible to what the composer had imagined in his head. And that quest for perfection and desire to define the work brought some new wrinkles. With the ability to seamlessly splice magnetic tape came the ability to correct mistakes, in both vocal and orchestral performances. When Wilhelm Furtwangler (1886-1953) conducted Kirsten Flagstad (1895-1962) in "Tristan und Isolde" in 1951, Flagstad could no longer reach some of the exposed high notes in the role; Elisabeth Schwarzkopf (1915-2006), the wife of the recording's producer, Walter Legge (1906-1979), was spliced in singing them for her. Once stereo arrived (in the shops by 1958) microphone technique had made great advances. At first only 2 microphones were used, but within a few years recording studios bristled with dozens, allowing orchestral details to be displayed as nobody but conductors and second violinists had ever heard them before, and manipulated with great virtuosity. Soon every section of the orchestra had its own microphone (or 3) and every soloist her own track, all of them fed by the dozens into huge mixing consoles. If the baritone was too loud in the quartet, the volume of his track could be lowered to please the tenor. If a singer couldn't hit that high C on the particular day the big aria was being recorded, it could be recorded again and again until she did hit it. And if she still couldn't hit it after many takes, her best effort could be speeded up very slightly to make up the difference. If she couldn't produce the diminuendo called for in the score, the engineers could lower the volume of her track to give the impression that she had. And eventually, if Mr. Domingo or Mr. Pavarotti couldn't be present in Paris during September to record his part of duets and ensembles, he could be recorded later—sometimes years later—and that track seamlessly integrated into the master tape. But is the result "a performance" at all? By the height of the digital era, when CD's were selling in large numbers and several dozen operas were being released every year, a typical "performance" might be assembled from literally thousands of very brief takes recorded over many months. Hearing was no longer believing.

Of course, singers really don't sound in the theater just as they do on recordings. Many smallish voices carry surprisingly well, but some get lost altogether; on records it's very hard to tell. Dramatic voices are harder to record than lyrics, so some artists get ignored by record companies; Leonie Rysanek (1926-1998), one of the most important sopranos of the post war period, made few commercial recordings; she was nearly impossible to record because her top sounded as loud as the whole orchestra. Birgit Nilsson (1918-2005) recorded frequently, but few of those recordings really capture the visceral excitement of her amazing high notes. The recordings Kirsten Flagstad made in her prime are impressive, but not quite like what an old timer told me about her: in the house, he claimed, her high notes were "like getting hit in the face with a baseball bat." Boris Christoff (1914-1993) is reputed to have had a rather small voice; on records he sounds enormous. And in general, almost all voices sound larger on records than the actual experience of hearing them in the theater. They tend to be miked more closely, so many record collectors (myself occasionally included) prefer to hear opera on records instead of in the opera house.

Of course, many live performances, largely airchecks, have survived from the time the microphone was introduced around 1925. Dozens of singers who made few or no commercial recordings have been preserved, Leyla Gencer (1928-2008) for example, an important artist who had a major career but whose legacy would be lost entirely were it not for her dozens of complete broadcast performances. Without the many great live performances of Maria Callas (1923-1977), we would hardly know what magic she made in live performance, no matter how wonderful her earlier studio recordings unquestionably are. Gertrude Grob-Prandl (1917-1995), an important post-war Wagnerian comes to mind, as does Astrid Varnay (1918-2006), who made more than a few commercial recordings, but whose live performances, especially from Bayreuth, are revelatory.

What we hear in these live recordings is not an abstraction or a musical mosaic, but, for better or worse, an evening in the theater. Stage noises, mistakes, hoarseness, flat high notes (or sharp ones), all manner of blunders abound. With DVD's we get absurd sets, unflattering costumes, directorial conceits, a good look at dental work, and often the rueful awareness that Madame X looks nothing at all like Mimi or Tosca, or Herr Z is much too old and stout to make anything like a passable (visual) Siegfried. But we get a real performance. With studio recordings, we get something more, and less.