[make_voxceleb1.pl] Fix train/test data split for the latest version of the voxceleb1 dataset

[sunshines14]

I found that previous scripts do not work with the latest version of the voxceleb1 dataset.
Therefore, I fixed the script for the latest version as follows:

1. 'url' update to download the latest version (from official site)
2. Delete the script about 'vox1_meta.csv' no longer needed
3. Fix the code for the latest version's output (variable name equals make_voxceleb2.pl)

It is confirmed that successful output can be obtained from the script.
ex) voxceleb1_test -- spk2utt, trials, utt2spk, wav.scp

[sunshines14]

(+) if previous script is still needed, I can use this modified script to another version (ex. v2) rather than replacing the existing script.

[danpovey]

Thanks a lot!! @david-ryan-snyder can you please check this?

[david-ryan-snyder]

Thanks @sunshines14.

Could you do what you offered and copy this to a new script called make_voxceleb1_v2.pl?

Then, in the run.sh script, make the v2 script the default one. Comment out the the old version of the script, and above it write a short comment describing the situation. E.g., mention that if you downloaded the dataset soon after it was released, you will want to use the make_voxceleb1.pl script instead.

[sunshines14]

Thanks @david-ryan-snyder.

I fixed some code as you commented as follows:

1. I have made a new script called 'make_voxceleb1_v2.pl'.
2. In the run.sh script, the v2 script used as the default one.
3. In addition, I have made the code simpler in 'make_voxceleb1_v2.pl'.

It was reconfirmed that successful output can be obtained from all of fixed scripts.
Thanks.

[david-ryan-snyder]

Thanks @sunshines14! I just suggested you credit your work in the v2 perl script and fix a preexisting typo. Then we can merge it.

[sunshines14]

I did it all.
Thanks @david-ryan-snyder.

[david-ryan-snyder]

Thanks @sunshines14, looks good to me. @danpovey, I think it's fine to merge this.