nsynthinstument in WebDataset Format

This dataset is the nsynthinstument dataset, formatted in the WebDataset format. WebDataset files are essentially tar archives, where each example in the dataset is represented by a pair of files: a WAV audio file and a corresponding JSON metadata file. The JSON file contains the class label and other relevant information for that particular audio sample.

$ tar tvf nsynth_train_sub55k_0000000.tar|head
-r--r--r-- bigdata/bigdata 365 2025-01-17 10:09 brass_acoustic_049-068-100.json
-r--r--r-- bigdata/bigdata 128044 2025-01-17 10:09 brass_acoustic_049-068-100.wav
-r--r--r-- bigdata/bigdata 355 2025-01-17 10:09 brass_acoustic_060-032-050.json
-r--r--r-- bigdata/bigdata 128044 2025-01-17 10:09 brass_acoustic_060-032-050.wav
-r--r--r-- bigdata/bigdata 364 2025-01-17 10:09 brass_acoustic_054-066-075.json
-r--r--r-- bigdata/bigdata 128044 2025-01-17 10:09 brass_acoustic_054-066-075.wav
-r--r--r-- bigdata/bigdata 355 2025-01-17 10:09 bass_synthetic_022-088-075.json
-r--r--r-- bigdata/bigdata 128044 2025-01-17 10:09 bass_synthetic_022-088-075.wav
-r--r--r-- bigdata/bigdata 356 2025-01-17 10:09 brass_acoustic_032-069-025.json
-r--r--r-- bigdata/bigdata 128044 2025-01-17 10:09 brass_acoustic_032-069-025.wav

$ cat brass_acoustic_049-068-100.json
{"note": 88873, "sample_rate": 16000, "pitch": 68, "instrument_source": 0, "instrument_family_str": "brass", "instrument_str": "brass_acoustic_049", "note_str": "brass_acoustic_049-068-100", "qualities_str": "['bright']", "instrument_source_str": "acoustic", "velocity": 100, "instrument_family": 1, "instrument": 437, "qualities": "[1, 0, 0, 0, 0, 0, 0, 0, 0, 0]"}