The representation of performance within a musical narrative such as opera or song is a compelling phenomenon whose study clearly has an important role to play in any understanding of the complicated nature of voice in music. Temporarily, music ceases to act as a narrative or expressive tool, and instead becomes a sonorous image recognized and perceived by those within that narrative. Rather than thinking or directly expressing themselves through song, characters quote song, as if hearing it in the same way that the audience does. Not surprisingly, such a phenomenon has proved the starting point for critical studies of voice and narrative; in just such a study, Carolyn Abbate begins with an examination of the ‘Bell song’ from Delibes's Lakmé, a narration by the opera's title character, whose ‘fetishization of voice as pure sound’ creates a ‘phenomenal performance, which might be loosely defined as a musical or vocal performance that declares itself openly, singing that is heard by its singer, the auditors on stage, and understood as “music that they (too) hear” by us, the theater audience’.

In another study of performance within opera, however, Edward Cone questions the distinction between what he describes as ‘realistic’ song (understood to be heard by the other characters within the drama) and ‘expressive’ song (portraying the character's thoughts or emotional state) by citing instances in the first act of Verdi's La Traviata in which music that initially works as the former subsequently returns as the latter.