This work aims to improve sound quality in the architectural project. Architectural practices are mainly dominated by the visual modality. Architects need tools to develop new projects considering sound criteria. Our purpose is basically beyond regulations and aims to manage sound phenomena when they are not necessarily synonymous of noise and annoyances. In other words, we can not reduce all this complexity and richness of the sound world in a simple problematic which compares noise with silence. The main hypothesis of this thesis is : the sound quality deals with the interactions between sensitive phenomena (perceived sounds) and people's activities within the space. Sound quality indicates certain qualities of the relations between sounds, space and social practices. Sound quality is not a fixed criterion of the environment. It embodies differently according to people and time. Consequently, sound quality of a space involves a crossed analysis between space, acoustics and human behaviours in situ. This research was supported by surveys which were carried out in three railway stations in Paris - Gare du Nord, Gare Montparnasse and Gare Haussmann (Eole line). Several walks inside each station have been chosen to value architectural devices, sound sources and observable social practices. The same action was studied (1) in all the three stations (2) at different moments during the week -with or without public in the space - (3) for both directions of the walk - go and return. Architectural analyses, sociological interviews and room acoustic criteria measurements have been performed in each station. This method allows us to show how sound quality perceived by users, depends on the co-existence of sociological, spatial and acoustical variables, as well as the ways by which the same space can create several sound qualities. All these results constitute a supply to the general theme concerning the predictability of sound quality in public indoor spaces.
