search

mazzzystar / randomCNN-voice-transfer

Audio style transfer with shallow random parameters CNN.
https://soundcloud.com/mazzzystar/sets/speech-conversion-sample
405 stars 75 forks source link

RuntimeError: CUDA out of memory #8

Closed ghost closed 5 years ago

ghost commented 5 years ago

Traceback (most recent call last): File "train.py", line 75, in content_loss = content_param compute_content_loss(a_C, a_G) File "/content/randomCNN-voice-transfer/utils.py", line 64, in compute_content_loss J_content = 1.0 / (4 m n_C n_H n_W) torch.sum((a_C_unrolled - a_G_unrolled) ** 2) RuntimeError: CUDA out of memory. Tried to allocate 2.99 GiB (GPU 0; 14.73 GiB total capacity; 10.14 GiB already allocated; 1.89 GiB free; 1.97 GiB cached)

mazzzystar commented 5 years ago

Sorry for the late response. It means your CUDA out of memory. You may try with shorter audio inputs, or change the N_FFT to get shorter spectrogram.