Sonidos sintéticos e falas humanas: a IA e a direção de atores na audioficção

Abstract

This descriptive report presents an experimental and practical study investigating the relationship between actor direction, sound-based artificial intelligence, and voice cloning in fictional audio productions. The research is structured around four main axes. The first analyzes the evolution of audio formats, tracing the historical trajectory from radio dramas to podcasts, with a focus on technological, aesthetic, and listening changes. The second discusses actor direction and its techniques, exploring methods for constructing vocal performance. The third addresses vocal expressivity, listening processes, and vocal Viewpoints as tools for creation and stage presence in audio acting. The fourth axis examines the uses of artificial intelligence and voice cloning, reflecting not only on their expressive limits and professional impacts but also on ethical issues such as voice misuse, risks of professional replacement, and biases embedded in vocal generation algorithms. The practical component consists of creating two episodes from the same script: one directed with human performance and another using AI-cloned voices. The comparison between these allows observation of the effects of direction on performance. A third episode compiles the research results, including interviews with voice and technology specialists as well as non-expert listeners, broadening the reflection on differences between human acting and artificial simulation.