PyTorch implementation of DeepWave

[iranroman]

Hello DeepWave,

Thank you very much for developing this model. It is truly amazing work that combines generative modeling with graph neural networks and acoustic imaging.

This is not really and issue, but I wanted to bring your attention to the pytorch implementation of your model that @adrianSRoman and I have been working on. Feel free to check it out, give us feedback, and share.

https://github.com/adrianSRoman/DeepWaveTorch

We will use it to carry out more experiments and build upon your work.

Cheers!

[matthieumeo]

Hi @iranroman, hi @adrianSRoman,

Thank you very much for your message, we were not aware of this effort of yours! We are very happy to know that you found our work interesting, and even more that you decided to extend upon it with a very much needed and useful PyTorch porting of the code base. We'll have a closer look at your work soon and will make sure to refer to your work on our repo too. Just out of curiosity, can you tell us in which context you are using the technique? We'd love to know more about your specific setup/application. Any paper you could point us to?
Also, we are curious about potential new features you may have in store for DeepWave (we have been meaning to come back on this work for a while, and have some ideas of improvement ourselves).

Cheers!

[iranroman]

Hello,

Thanks for your nice message. We do not have a paper yet. We are currently designing experiments to elaborate on your work. We will keep you posted as we make progress, but some of the things we want to do go from "low-hanging-fruit" such as adding recurrence over time, to other ideas such as adding "semantics" to DeepWave's acoustic imaging "representation" so that it also captures information about "what" is making a sound, not just "where".