Sound-to-Imagination: An Exploratory Study on Unsupervised Crossmodal Translation Using Diverse Audiovisual Data
This addresses the problem of enabling visual inference from sound for applications like assistive technology, though it is incremental with a focus on unsupervised methods.
The study tackled unsupervised sound-to-image translation using diverse audiovisual data without class supervision, achieving over 14% interpretable and semantically coherent images from unknown sounds on average.
The motivation of our research is to explore the possibilities of automatic sound-to-image (S2I) translation for enabling a human receiver to visually infer the occurrence of sound related events. We expect the computer to 'imagine' the scene from the captured sound, generating original images that picture the sound emitting source. Previous studies on similar topics opted for simplified approaches using data with low content diversity and/or sound class supervision. Differently, we propose to perform unsupervised S2I translation using thousands of distinct and unknown scenes, with slightly pre-cleaned data, just enough to guarantee aural-visual semantic coherence. To that end, we employ conditional generative adversarial networks (GANs) with a deep densely connected generator. Additionally, we present a solution using informativity classifiers to perform quantitative evaluation of the generated images. This enabled us to analyze the influence of network bottleneck variation over the translation, observing a potential trade-off between informativity and pixel space convergence. Despite the complexity of the specified S2I translation task, we were able to generalize the model enough to obtain more than 14%, in average, of interpretable and semantically coherent images translated from unknown sounds.