Zhenyu Tang

> torch-audiomentations I think (hope) I did those augmentations on the GPU since the incoming data is already on CUDA. I wrapped torch-audiomentations functions in a preprocessor class that was...

> > torch-audiomentations > > I think (hope) I did those augmentations on the GPU since the incoming data is already on CUDA. I wrapped torch-audiomentations functions in a preprocessor...

Your observation is correct - a random noise is used to generate phase. This is far from being ideal indeed. The real underlying issue is we don't have a good...

> > Your observation is correct - a random noise is used to generate phase. This is far from being ideal indeed. The real underlying issue is we don't have...

> This validation is carried out on cache files only, mostly to ensure that the initial download worked correctly. It gives better error messages. > > I'll see whether I...

> Sure! Here is my `dataset` and `dataloader` > > ``` > self.dataset = self.dataset.shuffle(1000).decode('pil').to_tuple("groundlevel.jpg", "overhead.jpg", "metadata.json","__key__").map(self.do_transforms).batched(self.args.train_batch_size).with_epoch(self.args.train_epoch_length) > > trainloader = wds.WebLoader(self.trainset, batch_size=None, > shuffle=False, pin_memory=True, num_workers=self.hparams.num_workers) > trainloader =...

It is possible. GSound has well-defined directivity described here https://github.com/GAMMA-UMD/pygsound/blob/8f41cb13da5dba9aa09cac3f42e668005ed5cf11/src/GSound/gsound/gsSoundDirectivity.h#L70-L83 But setting that flag to True only initializes an omnidirectional pattern which doesn't do anything, a specific pattern needs to...

> By the way, I just found that I fail to add a sound pattern to the source. I initialise the SoundSource with the pattern defined in the SoundDirectivity in...