🤗 Datasets adds audio and image documentation with Quickstart and ImageFolder guides

AI Impact Summary

With these updates, 🤗 Datasets extends support to audio and image modalities through dedicated guides and a revamped Quickstart that shows end-to-end loading and processing for PyTorch or TensorFlow. The Audio feature's automatic decoding/resampling and the ImageFolder dataset builder reduce manual coding, while to_tf_dataset simplifies integration with tf.data.Dataset pipelines. This accelerates onboarding and data prep for multimodal models, enabling faster experiments with audio and image datasets like GigaSpeech. Acknowledging future expansion, AudioFolder is teased as coming soon, signaling ongoing diversification of supported modalities.