About WaveAutomaton

WaveAutomaton is an independent research lab focused on building machine learning systems for understanding and modeling real-world audio.

We explore how raw acoustic signals can be transformed into structured representations that enable robust inference, classification, and prediction in complex environments.

Our work combines classical signal processing with modern deep learning architectures.

We apply techniques such as spectral analysis and MFCC feature extraction alongside convolutional, recurrent, and transformer-based models to build systems that operate across time, frequency, and context.

We investigate multimodal learning where audio interacts with vision and language, enabling richer semantic understanding of real-world data.

The focus is not only on model design, but also on the engineering of reproducible experimental systems for training, evaluation, and benchmarking.