We have experienced frequent failures during download of the large and all reazonspeech dataset, seemingly due to network issues (such as TimeOutErrors).
This was also experienced by the kotoba-tech team, described here. They created a manual downloader to try to sidestep this issue.
We should perhaps include scripts to download the dataset into smaller pieces, or prepare smaller chunks of the dataset for users to download to reduce the likelihood of failures during download of the larger splits of the dataset.
We have experienced frequent failures during download of the
largeandallreazonspeech dataset, seemingly due to network issues (such as TimeOutErrors).This was also experienced by the
kotoba-techteam, described here. They created a manual downloader to try to sidestep this issue.We should perhaps include scripts to download the dataset into smaller pieces, or prepare smaller chunks of the dataset for users to download to reduce the likelihood of failures during download of the larger splits of the dataset.