My pianist Isaac Cohen and I recently performed the Prokofiev Flute Sonata at the First Unitarian Church of Chicago. The church has a generous, echoing acoustic. It is beautiful to listen to, though not always the easiest space in which to play—particularly for the flute. To project into the space you must give the instrument a great deal of breath.

Prokofiev was famously dry, terse, allergic to sentimentality. He probably just wrote this sonata because the flute is an underrated instrument and he wanted to do something about that — thus a piece that is so technically demanding and classic Prokofiev — precussive piano lines, mysteriously beautiful chord progressions, and crazy scales up and down as if flutes don't need to breathe.

I think I played reasonably well. Yet after the performance I felt terrible. In my memory the piece seemed full of mistakes—slips, cracks, moments where the line did not quite bloom the way I imagined it.

It reminded me of a study:

The Art of Inaccuracy: Why Pianists’ Errors Are Difficult to Hear

Bruno H. Repp | Music Perception (1996) | https://doi.org/10.2307/40285716

…Most errors seemed likely to be perceptually inconspicuous. This was confirmed in an error-detection experiment, in which eight pianists, some of whom had recently studied the test piece (the Chopin prelude), collectively detected only 38% of all objectively registered errors. Pitch errors, rather than being a categorical phenomenon (as a scorebased analysis might suggest), vary in the degree to which they violate the music, and their perceptibility is context-, listener-, and situation dependent. Members of a typical concert audience are likely to notice only a small fraction of a pianist’s inaccuracies, which is in part due to the contextual appropriateness of most errors.

Well, fairly speaking the piano is in some sense protected by its own richness. It is a polyphonic instrument, and when sound blends in a resonant space you rarely hear every individual note with equal clarity. The texture itself offers a kind of camouflage. In live performance there is also a psychological softness.

That famous 38% detection rate is probably a lower bound.

Woodwinds, by contrast, live a slightly harder mode. Our line is exposed. A single pitch occupies the air, and when listeners know the repertoire well they can often hear even small imperfections—tiny instabilities of pitch, a note placed a shade too early or too late.

And yet the performance was warmly received. So perhaps the performer’s inner accounting of errors belongs mostly to the performer. Audiences listen for something else: shape, direction, breath, perhaps even the courage of the attempt.

Thought I would note that audience mistake-perception has never been part of a performer’s true optimization function. Still, it is occasionally comforting to remember that music, like life, is heard more generously than it is judged by the one who makes it.