search

lsfhuihuiff / MusicTI_AAAI2024

" Music Style Transfer with Time-Varying Inversion of Diffusion Models"
MIT License
35 stars 4 forks source link

Error 404 of Dataset. #3

Open spiralanch opened 9 months ago

spiralanch commented 9 months ago

Hi, @lsfhuihuiff I am trying to run training, but met multiple problems below.

  1. Dataset is 404 on ("https://lsfhuihuiff.github.io/MusicTI/"), is it possible to make it public?
  2. Training dataset is not well understand in paper, how can i prepare style audio and content audio, then retrain the model? Could you please explain the strategy of the dataset?

Thanks in advance!

lsfhuihuiff commented 9 months ago

You can start by experimenting with the data samples in the "audios" directory. All the audio data has been segmented into 5.12 seconds. Convert them into mel-spectrograms and use them for training and inference.