Andrew Harding

Title: Andrew Harding and Audio Generation Research.

Introduction

Andrew Harding is an inventor based in Sherman Oaks, California. His patent work addresses how audio data can be modified to affect the training of generative machine-learning models.

Latest Patents

Harding is associated with 1 patent, titled “Method to Disrupt Generation Quality of Deep Audio Generation Models.” The method segments an audio signal and uses a trained machine-learning model to create modified segments. These segments are designed to sound indistinguishable from the originals to the average human listener. After the modified segments are combined into a reconstructed signal, that signal is likewise intended to sound indistinguishable from the original. However, using it to train a generative model constrains the model’s ability to generate new audio similar to the source signal.

Career Highlights

Harding is associated with Sourceaudio Holdings, LLC. His listed patent reflects work at the intersection of audio processing, machine learning, and techniques intended to influence the behavior of generative models.

Collaborations

The provided information lists Michalis Papakostas and Brent Phillips as Harding’s coworkers. It does not specify their roles in the patent or the extent of their collaboration.

Conclusion

Andrew Harding’s patent describes an approach that preserves the perceived sound of audio while limiting its usefulness for training models to generate similar audio.