Kun Zhou

icon indicating a website link https://kunzhou9646.github.io/ icon indicating an email [email protected]

icon company Human Language Technology Lab, NUS icon location Singapore icon location PhD student in National University of Singapore (NUS).

Results 3 issues of Kun Zhou

[Feature]: Audiocap dataset dev and test files

2 comment

Thanks for sharing the Audiocap dataset. However, I note that the audiocap provided only contains training files. Can you also share dev and test files? Thanks a lot!

enhancement

Data Preprocessing

Thanks for your implementation! I am curious if you could share data preprocessing scripts to get "audio_ann_sum.txt" and "audio_sum.hdf5" on mixed of Libritts and aishell?

[Help]: Reproduced TTA Results

I reproduced TTA results accordingly, however, they are far from the performance that reported in the original AUDIT paper. I wonder if there is any pre-trained models or demos from...