The works of well-known composers active as recording technology developed and as the recording industry emerged thus make ideal case studies. Giacomo Puccini is uniquely suited to a study of tradition through technological means: he has perhaps the best-documented relationship to advances in technology and the resulting shift in entertainment aesthetics of any composer of this era. Of his twelve operas, Turandot is the only major work whose premiere post-dates the advent of electronically captured and controlled sound recording, and thus position at the crossroads of technical practices and social aesthetics stemming from technological conditions in multimedia entertainment in the 1920s.

Of the major opera houses in the world, New York’s Metropolitan Opera has perhaps the most comprehensively documented performance history. The launch of “The Metropolitan Opera Archives” online with open access in 2005 made it the most accessible source of information for operatic performing history in any opera house in the world. By making this information publicly and readily available, the Met made significant strides in democratizing opera in a way similar to what recordings had done nearly a century earlier. With this new tool, fans now had the ability to deepen their learning and, because of this, their enjoyment of opera at the Met. Thus, by extending the social activity of learning about opera audiences to the digital (and therefore widely accessible) realm, the Met sought to thrust itself into the forefront of the operatic social sphere in addition to its broadcast offerings. Many recordings of these transmissions are available online through The Metropolitan Opera On Demand, while the Rodgers and Hammerstein Archive of Recorded Sound at the New York Public Library serves as the official repository for hard and digital copy recordings of the broadcasts.

Within this opera, six moments are most appropriate for examination for both practical and dramaturgical reasons. Arias in Puccini’s operas offer dramatic pauses for characters; they are moments of emotional exposition, or the result of reflecting on surrounding events. Both of these types of arias occur in Turandot. Liù’s first aria, “Signore, ascolta!”, exhibits both functions, as she reacts emotionally to Calaf’s stated intent to answer Turandot’s riddles. Calaf’s response, “Non piangere, Liù,” gives him the opportunity to respond tenderly to Liù, and show concern for his father, Timur, by requesting she remain if the prince surrenders his life. Turandot’s entrance aria (in Act II), “In questa reggia,” recounts the reasons for her riddles and brutality – her desire for vengeance on behalf of her ancestress Lou-o-Ling. “Nessun dorma!” is Calaf’s reflection on Turandot’s commandment that no one in the realm shall sleep until she knows his name and shows his resolve in conquering the princess. Liù’s suicide aria, “Tu che di gel sei cinta,” reveals both her crumbling determination under the duress of torture and her desire for Calaf to win the hand of Turandot. In addition to these arias, one other moment of musical practicality and dramaturgical significance merit consideration for its performance traditions: the dramatic climax of acts II. In Act II’s "Riddle scene," where Turandot poses each of her three riddles and Calaf successfully answers them, fermatas precede the statement of each riddle and Calaf’s response to the first and third. The time performers devote to observing these fermatas in relationship to the surrounding musical fabric can indicate varying degrees of gravity, levity, or expediency as the opera’s second dramatic conflict emerges from the resolution of its first.

Both digital (CD and DVD) and analog (cassette or reel tape) recordings of Turandot comprise the available corpus as of 2016. Some of these items are commercially available through legitimate means while others are only accessible at NYPL. Bootleg recordings are available for all others identified here, and many other recordings in Turandot’s performance history, both for sale or via internet streaming or download. This dataset comprises only legally accessible recordings.

Commercially available recording data extraction

For recordings that are commercially available, either in digital form or able to be converted into it, the data generation method uses the same approach as the Mazurka Project (http://www.mazurka.org.uk/). Upon loading the file into Sonic Visualiser, beat entry occurred with reference to a score. Beat markers were moved as necessary, and then the onset detector plugin was applied to refine marker positioning. Tempo-to-following-beat data extraction followed.

Archive-only available recordings data extraction

Copyrights prohibit any duplication of recordings not already commercially available, thus alternative means of generating tempo data are necessary. For this process, I adopted principles of aural dictation. First, I listened to a recording and made annotations in a score to reflect places of tempo stasis and tempo modulation. On a second listening, I created a tempo-tap-track using the voice notes recorder on a cellular phone. The next step was to listen to the tap track in one ear and the aria recording in the other, making further notes in the score of passages that were accurate and others that needed adjustment. This refining process repeated several times. Upon loading these audio files into Sonic Visualiser, all tap tracks were compared against each other, and in reference to score annotations, either the most accurate tap track section’s beat marker was used, or, in isolated instances, a beat marker was assigned to the average of the closest tap tracks. While obviously imperfect, this is one of the only means available for generating data from copyright protected recordings. Furthermore, any discrepancies that do exist between final beat marker assignment and actual beat placement within the current data set are within hundredths, if not thousandths of a second. The same method of extracting tempo data was then applied to these tap tracks as to recordings with digital copies commercially available.

Additional Data

Each aria has one .csv file accompanying it, divided into two blocks separated by two blank columns. The first block contains timestamp and tempo data. The second contains a moving or progressive average, which represents one way of sketching the notion of performance tradition in empirical terms.

As each of the six individual single solo-passage excerpts of the opera, examination/analysis of the evolutionary nature of a performance history is perhaps an obvious way to analyze the data.

If one is interested in the 'large-block' pacing as it flows throughout the opera, the order in which the excerpts appear is: "Signore, ascolta!", "Non piangere, Liù," "In questa reggia," "Riddle Scene (Straniero, ascolta!)", "Nessun dorma," and "Tu che di gel sei cinta."

If one is interested in looking at individual characters excerpts, the pairings are: "Signore, ascolta!" & "Tu che di gel sei cinta" for the slave girl Liù; "Non piangere, Liù" & "Nessun dorma" for the prince Calaf; and "In questa reggia" and the "Riddle Scene" ("Straniero, ascolta!) for the princess Turandot.

Text has also been provided on a beat-by-beat breakdown to enable consideration of textual relationships to tempo and could be a fruitful area for exploration.

Also included is the performance history data including singers, conductors, chorus master, production designer, costume designer, choreographers, and house manager. For performances where no information is available via the Metropolitan Opera's online archive, (http://archives.metoperafamily.org/archives/frame.htm), 'none listed' appears in the table.

Each excerpt and the performance history are offered in both .csv and .xml format.