The tfrecord files about avsr-tf1 HOT 8 OPEN

Hi @xjwla

Thanks for opening the issue.
Could you please paste the error message here ?

On my local copy of the TCD-TIMIT dataset I made a few corrections, most of them for speaker 42M.
That script uses the fixed dataset partitioning, as listed here. I suspect that your download is not incomplete, and you only need to rename a few files.

It would not be possible for me to share raw or processed data because it is licensed.

from avsr-tf1.

HI,

I am also facing the same issue when I run write_records_tcd.py. I am able to run till I reach this error:

FileNotFoundError: [Errno 2] No such file or directory: './datasets/tcdtimit/volunteers/01M/Clips/straightcam/si2077.wav'

Am I suppose to download a local copy of the TCD-TMIT dataset before running the script?

Thank you in advance for your help!

from avsr-tf1.

@clarahohohoho
Yes, you would have to manually download any dataset you would like to use this project with.
The script is only meant to serve as an example for writing your data into a format compatible with the code.

Most datasets that I've worked with required me to sign up with an academic email or fill in license agreements. I did not see an easy way to automate this process. After all, data preparation is still one of the most time consuming aspects of machine learning.

from avsr-tf1.

Hi George,

Appreciate your quick reply on this! Unfortunately while working on this, the download link provided on (http://www.mee.tcd.ie/~sigmedia/Resources/TCD-TIMIT) is unavailable at the moment due to the upgrading of the servers. Are you able to share the raw TCD-TIMIT dataset with me?

Thank you!

from avsr-tf1.

@clarahohohoho
Sorry, I am not aware of an alternative download page for TCD-TIMIT.
I contacted the administrator of the webpage you mentioned. They are still working on finding a new server.

This github repository is not related to the management of the TCD-TIMIT dataset, and I recommend that you contact my supervisor for any further assistance on this topic.

from avsr-tf1.

Sorry, another question, I didn't know which is the true label_file in write_records.py for LRS2 dataset. I couldn't find out the file "contain pairs of (example name - transcription) on each line, delimited by a space" in the whole LRS2 dataset. If something is wrong with my understanding, please pardon me.

Thank you!

from avsr-tf1.

@ZhangYX-bin
Sorry, I missed your issue on the Taris repository. Will reply there.

from avsr-tf1.

Hi George,

Appreciate your quick reply on this! Unfortunately while working on this, the download link provided on (http://www.mee.tcd.ie/~sigmedia/Resources/TCD-TIMIT) is unavailable at the moment due to the upgrading of the servers. Are you able to share the raw TCD-TIMIT dataset with me?

Thank you!

Hi, Unfortunately, I had the same problem that the download link provided on (http://www.mee.tcd.ie/~sigmedia/Resources/TCD-TIMIT) is unavailable. May I ask if you have found a solution. Thank you very much

from avsr-tf1.