Bakhtin defines the dialogical text as an "arena of conflict between two voices." Mary Ann Doane expands this concept to distinguish monological from dialogical texts in film. A monological text often conceals its voice in order to sustain the spectator's identification with the fiction, while a dialogical text manifests multiple voices to create manifold identifications. In a dialogical text, dispersed voice-over narration often disrupts the unisonic discoursing voice. In The Acoustic Mirror, Kaja Silverman suggests that in the classical Hollywood cinema, the soundtrack is organized by sexual difference. Male voice-over tends to be synchronized with visual track and associated with exteriority, that is, in the framing of the diegesis. Pointing to the fact that female non-diegetic narrators are virtually non-existent in American feature films, Silverman argues that the male voice enjoys a privileged position in the text and is structured as the site of enunciation. Female voice-over, on the other hand, is temporally and spatially dislocated from the image track and associated with interiority. It is invariably confined by the diegesis, and thus its position relative to the masculine voice is restricted from enuciative power.
Since City of Sadness' female voice-over and female writing are closely linked in the diegesis, feminine voice and its relation with masculine voice in creating dialogic qualities will be discussed in the section on writing in a more cross-referenced analysis. This scope of this section primarily to the soundtrack itself. The diegetic sound in City of Sadness is organized in a polyvalent style and for the purposes of simplicity we cleave our discussion into two sections in accordance with sexual difference: feminine voice and masculine voice.