Generative adversarial networks to synthesise audio data for building underwater acoustic data
- Generative Adversarial Networks are a highly powerful tool in machine learning and can be used to create large quantities of data.
- GANs have been used to create audio data in other disciplines, such as speech, and this data can be highly accurate and realistic.
- This paper will demonstrate how GANs can be used in the field of underwater acoustics to create realistic synthetic acoustic data in an area where large amounts of data is difficult to procure. These data can then be used in many areas of further development from sonar algorithm design and to realistic modelling of a sonars capability in certain environments.
Chairperson
