Total Hours
#2
by
greyfaulkenberry
- opened
Thank you so much for this dataset! It will really be a boon for testing and training. I had a quick question about total duration. In the paper High-Precision Medical Speech Recognition Through Synthetic Data and Semantic Correction: United-MedASR it says there were ~790,000 files, at 30 seconds each, totaling 5,486 hours of labelled audio. I may be missing something, but when I examined the dataset, most of the files are less than 30 seconds, so I assume this is a reduced version. Are there any plans to release the full version at some point in the future?