This dissertation is about audio description (AD), a subject included in the field of audiovisual translation (AVT), more specifically in the area of intersemiotic translation. AD is the transformation of image into verbal text, i.e., of visual code into auditory code, and is performed on cultural products in order to make them accessible to visually impaired people, a spectrum that includes congenital blindness, acquired blindness and low vision. Even though AD can occur in different products, this study will be limited to motion pictures, in two different dimensions — one fundamentally theoretical and the other practical — that are interrelated and influence each other. In the first dimension, a discussion about interpreting and describing is triggered by the requirement of limiting interpretation to a minimum. Norms in general establish the search for objectivity in order to prevent manipulation or paternalistic attitudes, thus avoiding subjective choices, usually related to the use of qualifiers. Frequently, both lay people and specialists still conceive describing and interpreting as a dichotomy, and prioritize description, seen as objective, over interpretation, considered as subjective and valuational. The study aims at deconstructing this dichotomy by means of a many-valued logic that acknowledges the empirical need for the terms description and interpretation, but criticizes the absolutizing and excluding senses attributed to them. In the second, more empirical dimension, audio descriptions of characters gestures — understood as mimicry and emotional states — were subjected to reception analysis. Gestures were chosen because there has been little or no study at all about them and because they are quite convenient when it comes to reflecting upon the recommendations of not interpreting or of limiting interpretation to a minimum. The research aimed at determining whether more interpretive audio descriptions may be necessary, or even indispensable, to the enjoyment of motion pictures, and also whether, according to the type of visual impairment, there are different preferences for less or more descriptive or interpretive ADs. The study included two groups, one made up of students from the Benjamin Constant Institute (IBC) and another of members of the Associação dos Deficientes Visuais do Estado do Rio de Janeiro (Association of the Visually Impaired of Rio de Janeiro) (ADVERJ). The group of IBC students watched the film O palhaço (The clown), and the members of ADVERJ watched the film Menos que nada (Less than nothing); after that they were individually interviewed and answered a questionnaire with questions pertaining to the comprehension of the film itself and aspects specific to gestures. They also watched four passages of the movies again, one for each type of gesture (substitutive, divergent, simple and complex affect displays); each passage had two AD versions — one more descriptive and another more interpretive, with more adjectives and adverbs —, so that they could choose the version they preferred. The proposal, then, was to test theoretical conclusions in an empirical context of strictly visual information and to reflect upon AD characteristics that may be more adequate to Brazilian audiences in Rio de Janeiro, specially gesture AD. In its initial part, the dissertation presents a historiography of AD both as an activity and as an object of investigation. It also shows how AD came to be present in the different spheres of the media and the academy in Europe, the United States and Brazil.