PyTorch implementation for SoundNet paper link.[1] SoundNet is a model to classify sounds using transfer learning with visual knowledges on large size of unlabeled videos.
The original implementation by cvondrick link was using Torch. Alternative implementation with TensorFlow was created by eborboihuc link .
- PyTorch website
- NVIDIA GPU + CUDA 8 + CuDNN v5
- Python 2.7
.... To be Continued ....
Reference [1] SoundNet: Learning Sound Representations from Unlabeled Video : By Yusuf Aytar, Carl Vondrick, Antonio Torralba. NIPS 2016