Sumu Zhao

Hello, Gong. So you want to train the genre classifier? If you want to train cyclegan model, the type should be cyclegan instead of classifier. I think the shape of...

I tried by myself and it actually can run. I checked `batch_images = np.array(batch_images).astype(np.float32)`. The shape of batch_images is (16, 64, 84, 2). I think there might be something wrong...

Well, it's weird. It should run... Did you ever change something?

I checked the load_npy_data() function in the utlis.py. The shapes of npy_A, npy_B and npy_AB are as expected: (64, 84, 1), (64, 84, 1), (64, 84, 2). So I'm also...

I have uploaded the datasets. You can find it in the Dataset part in ReadMe.

Sorry I don't have time to do this recently due to the final exams. I'll do it later.

Hi, thanks. In MuseGAN paper, they sampled each bar by a sampling rate of 96 time steps. While in our case, each phrase is composed of four consecutive bars. Each...

No, I didn't use LPD or LMD. You can download via the link in Datasets part in ReadMe. Please read ReadME file carefully. Basically, everything has been introduced clearly. For...

I use the packages pretty_midi and pypianoroll a lot in my project. They are really nice tools to deal with MIDI files. I recommend you could read their documents and...

Hey, basically in my project, I just merge all the tracks into one single track. Multiple tracks would be way more difficult than single track. This is also what we...