https://github.com/harritaylor/torchvggish

Pytorch port of Google Research's VGGish model used for extracting audio features.
https://github.com/harritaylor/torchvggish

audio-embedding audioset pytorch vggish

Last synced: 17 days ago
JSON representation

Pytorch port of Google Research's VGGish model used for extracting audio features.

• Host: GitHub
• URL: https://github.com/harritaylor/torchvggish
• Owner: harritaylor
• License: apache-2.0
• Created: 2019-02-04T17:13:17.000Z (almost 6 years ago)
• Default Branch: master
• Last Pushed: 2021-11-03T10:38:12.000Z (about 3 years ago)
• Last Synced: 2024-10-22T19:41:01.159Z (22 days ago)
• Topics: audio-embedding, audioset, pytorch, vggish
• Language: Python
• Homepage:
• Size: 316 KB
• Stars: 377
• Watchers: 8
• Forks: 68
• Open Issues: 13
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
**Looking for maintainers** - I no longer have the capacity to maintain this project. If you would like to take over maintenence, please get in touch. I will either forward to your fork, or add you as a maintainer for the project. Thanks.

---

# VGGish

A `torch`-compatible port of [VGGish](https://github.com/tensorflow/models/tree/master/research/audioset)[1], 

a feature embedding frontend for audio classification models. The weights are ported directly from the tensorflow model, so embeddings created using `torchvggish` will be identical.

## Usage

```python

import torch

model = torch.hub.load('harritaylor/torchvggish', 'vggish')

model.eval()

# Download an example audio file

import urllib

url, filename = ("http://soundbible.com/grab.php?id=1698&type=wav", "bus_chatter.wav")

try: urllib.URLopener().retrieve(url, filename)

except: urllib.request.urlretrieve(url, filename)

model.forward(filename)

```



[1]  S. Hershey et al., ‘CNN Architectures for Large-Scale Audio Classification’,\

    in International Conference on Acoustics, Speech and Signal Processing (ICASSP),2017\

    Available: https://arxiv.org/abs/1609.09430, https://ai.google/research/pubs/pub45611