Datasets:
Issue with mismatched audio and image data
I am currently working with the VAANI dataset, and I have observed that the audio data and their associated images do not seem to be properly aligned. In almost all cases, the referenced image filenames do not correspond correctly to the audio samples. This leads to potential issues during model training.
I would appreciate it if you could confirm whether this is a known issue, and if there are any plans for releasing a corrected mapping or metadata file to resolve these inconsistencies.
Thank you for pointing this out. We have been updating the data by adding over 40 additional districts, which may have caused the issue. The data upload is now complete. Could you please check again? If the issue still persists, we would appreciate it if you could share some examples.