Custom Audio PyTorch Dataset with Torchaudio

In the video, you can learn how to create a custom audio dataset with PyTorch loading audio files with the torchaudio. As a use case, we'll be using the UrbanSound8K dataset.

In the process, you’ll also learn basic I/O functions in torchaudio.

Code:
github.com/musikalkemist/pytorchforaudio/tree/main…

UrbanSound8K dataset webpage:
urbansounddataset.weebly.com/urbansound8k.html

Torchaudio backend documentation:
pytorch.org/audio/stable/backend.html

===============================

Interested in hiring me as a consultant/freelancer?
valeriovelardo.com/

Join The Sound Of AI Slack community:
valeriovelardo.com/the-sound-of-ai-community/

Connect with Valerio on Linkedin:
www.linkedin.com/in/valeriovelardo

Follow Valerio on Facebook:
www.facebook.com/TheSoundOfAI

Follow Valerio on Twitter:
twitter.com/musikalkemist​

===============================

Content:
0:00 Intro
0:21 Urban Sound 8K Dataset webpage
1:14 PyTorch Dataset and DataLoader classes
2:02 UrbanSoundDataset class
4:35 Implementing the constructor - part 1
5:24 Content of Urban Sound 8K
7:33 Implementing the constructor - part 2
8:53 Implementing _len_
9:19 Implementing _getitem_ - part 1
11:28 Torchaudio backend I/O
12:34 Implementing _getitem_ - part 2
13:20 Getting audio sample path
17:06 Getting label
17:49 UrbanSoundDataset class recap
18:26 Running UrbanSoundDataset
21:53 Coming up next