HIN0209

Thank you for the response. I checked it and saw multiple potential solutions. Can you suggest what part of your code these should be inserted into? From a novice pythoner

Hi, the URL below provides wav and mel files, but I cannot find train.json and valid.json. Are we supposed to create by us, but no caption list is available? https://openxlab.org.cn/datasets/Amphion/AudioCaps/tree/main

Hi, using Jupyter lab, I was successful converting in a batch manner with the code below. All the wav files in the input_dir are converted to 5-second spectrograms. (reference: https://github.com/thx-pw/riffusion-music2music-colab)...

my prior code needs to be adjusted for indents for the last two lines, obviously.

AttributeError: module 'torch' has no attribute 'rfft' with PyTorch https://stackoverflow.com/questions/67647299/attributeerror-module-torch-has-no-attribute-rfft-with-pytorch use torch.fft.rfft instead of torch.rfft.

The 'learning_rate' of 0.003 might be too large, as written in the Hyperparameter tuning, I guess this is the first point to check. I am also curious about the size...

What was the matter with the following command? "--temp_ckpt_folder "temp"" does not work with me, such that I copy-paste the checkpoint-1254 (in this example) instead. python3 transcribe_audio.py \ --is_public_repo False...