Visual Audio: An Interactive Tool for Analyzing and Editing of Audio in the Spectrogram

We present a tool for analyzing and editing audio signals in the visual domain. As visual representation we use spectrograms, which give descriptive information about the sound. This allows analysing and editing audio in a “what you see is what you hear” style. Gabor analysis and synthesis serves as a basis to create images and recreate audio signals from the edited images in hi-fi quality. As the structures in the spectrogram are rather complex, image processing and computer vision methods are applied for smart user assisted editing. Templates based on sounds recorded under defined conditions are therefore used. This allows detecting, separating, eliminating and/or modifying audio objects supervised (interactively) or automatically. We further propose the usage of resolution zooming, to support manipulating the spectrogram of a signal at any chosen time-frequency resolution