After choosing the operas, finding arias and duets was actually quite simple. I decided that they would not have to show what my characters think or do, but rather, that the pieces would show release from the real world. So, actually, any aria or duet would do.
For me, however, it was important to take them apart, at least in a way.
For example, I took Violetta’s Scene “è strano… sempre libera” and only put some of the parts in the theatre. Also, I didn’t let those parts come one after the other and left the “ah forse lui” away.
Another example with Germont and Violetta’s scene: Although I started with “bella voi siete e giovine”, I didn’t let them finish it and went on very quickly with “non sapete quale affetto”.
Also, I made new duets by combining arias together – because the butler’s character would become increasingly drunk, he would forget one role and sing another instead, which would get on Miss Mary’s nerves, so she would sing in between. Thus, a combination of Leporello’s aria “notte e giorno faticar” and Rosina’s “io sono docile” was born (even though I knew it would be very difficult to sing it, I was confident we could).
OK. Ideas ready. Putting all in one document and make it clear to musicians was next. Scanning all music, taking it apart and putting it together was nothing but computer savviness. As said, it was done.
I still had to ‘compose’ the recitatives, though. Not really being a composer myself, this was another big challenge I had set for myself. The words were ready already, so it was just a matter of piano and voice.
How could I do this? What is important for ‘sung dialogue’?
Looking back, I ran (or bulldozed) right through the task. I decided to keep it as simple as possible with recitativo secco (à la Rossini) and keep the musicality of the pieces before and after, so changing it somewhere in between. For the voice I tried to sing it myself (even the butler) and decided on a spoken musicality with elements of our time.
Now we were ready for practice.