pytorch/audio

Deprecate data utils

Closed this issue · 3 comments

mthrok commented

PyTorch core has download function and torch.hub.download_url_to_file.
Torchaudio can use it for dataset download and does not need to maintain its own torchaudio.datasets.utils.download_url.

In addition to that, there seems to be something wrong with torchaudio.datasets.utils.download_url.
It fails to download multiple of huge files. I think we should get rid of torchaudio.datasets.utils.download_url and torchaudio.datasets.utils.stream_url.

Steps

Phase 1

  1. Replace the usages of torchaudio.datasets.utils.download_url in torchaudio with torch.hub.download_url_to_file and torchaudio.datasets.utils.validate_file
  2. Add deprecation warning to torchaudio.datasets.utils.download_url and torchaudio.datasets.utils.stream_url, saying it will be removed in v0.12.

Phase 2

Once the v0.11 is released and the main branch is bumped to v0.12, we can remove these functions.

Note

For setting up the development environment, please refer to CONTRIBUTING.md and torchaudio_unittest

jdsgomes commented

I am working on this issue

radioactive11 commented

@mthrok can I work on this issue?

mthrok commented

Hi @radioactive11

Sorry for the late reply. The phase 1 is completed and we need to wait until the branch cut of v0.11 to perform the phase 2. I will update this once the branch cut happens.

  • 👍1