Audio Classification

Classify the audios. In this repo, I train a model on UrbanSound8K dataset, and achieve about 80% accuracy on test dataset.

There is a pre-trained model in urban_sound_train, trained epoch is 1000

Usage

• audio_train.py: Train audio model from scratch or restore from checkpoint.
• audio_params.py: Configuration for training a model.
• audio_inference_demo.py: Demo for test the trained model.
• ./audio/*: Dependencies of training, model and datasets.
• ./vggish/*: Dependencies of VGGish for feature extracting.

Tools

• TensorFlow: VGGish
• Google AudioSet
  • VGGish model checkpoint
  • Embedding PCA parameters
• pyAudioAnalysis(Ref.)

Dataset

• urban sound dataset

Ref. Blogs

• AudioSet: An ontology and human-labelled dataset for audio events
• CNN Architectures for Large-Scale Audio Classification