Fine-tuned on classification datasets and Add augmentations

This release includes the following:

• Fine-tuned on ESC50, FSDKaggle2018, SpeechCommandsv1
• Add waveform augmentations
• Add spectrogram augmentations
• Add augmentation testing notebook
• Add tagging metrics

Bug fixes, add datasets and fine-tune tasks

Add the following datasets:

• ESC50
• UrbanSound8k
• FSDKaggle2018
• SpeechCommandsv1/v2

Release fine-tuned model on ESC50.