BrahmiVoice

BrahmiVoice is an exploratory multimodal AI project investigating the relationship between vowel acoustics and written-character geometry. The system extracts 31 acoustic features from isolated vowel recordings, including MFCCs and spectral characteristics. A locally trained model maps these features into a low-dimensional glyph representation, which is reconstructed into a two-dimensional Signed Distance Field (SDF). The project also integrates Meta Muse Spark to interpret predicted glyph geometry alongside acoustic features, helping identify possible visual patterns, discuss limitations, and suggest future experiments. BrahmiVoice investigates how much information about visual geometry may be recoverable from acoustic features alone. It is an exploratory prototype, not a claim that sound determines historical writing systems or that the generated shapes are historically authenticated characters.