{"success":true,"data":{"author":"Jesse Engel, Cinjon Resnick, Adam Roberts, Sander Dieleman, Douglas Eck, Karen Simonyan, Mohammad Norouzi (Google Magenta, DeepMind)","bucketAccessType":"public","bucketName":"jg583/NSynth","bucketPrefix":"","category":"Speech & Audio","cloudProvider":"huggingface","createdAt":"2026-11-14T14:15:17.993Z","description":"NSynth is an audio dataset of over 300,000 musical notes from 1,006 instruments from commercial sample libraries, released by Google Magenta to give audio machine learning a high-quality entry point comparable to MNIST or ImageNet. Each four-second, 16 kHz monophonic recording holds a note played for three seconds and left to decay for one second, across every pitch of a standard MIDI piano the instrument can produce at five velocities. Every note is annotated with its instrument source (acoustic, electronic or synthetic), instrument family (such as bass, brass, guitar, keyboard, mallet, organ, reed, string, vocal) and sonic qualities (bright, dark, distortion, percussive, reverb and others), from human evaluation and heuristics. It is a benchmark for neural audio synthesis (WaveNet autoencoders, GANSynth, DDSP), timbre representation learning and pitch and instrument classification. This repository is a Parquet conversion with train, validation and test folders under data/; train holds 35 Parquet batches of about 27.7 GB (measured 2026-11-14; the Hub reports about 29 GB). Load with datasets.load_dataset('jg583/NSynth') or pyarrow. Google released NSynth under CC BY 4.0; cite Engel et al., 'Neural Audio Synthesis of Musical Notes with WaveNet Autoencoders', ICML 2017.","displayTitle":"NSynth: Neural Audio Synthesis musical notes dataset","id":"676a3dd0-9bba-5183-8df1-eee9e4d98a70","license":"CC-BY-4.0","licenseEvidenceUrl":"https://magenta.tensorflow.org/datasets/nsynth","objectCountEstimate":40,"probeStatus":"ok","referenceUrl":"https://arxiv.org/abs/1704.01279","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"jg583/NSynth\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"jg583/NSynth\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=jg583, license:cc-by-4.0, size_categories:100K<n<1M, downloads=2361"],"format":"mixed","generatedAt":"2026-11-14T14:15:20.871Z","layout":{"example":"data/test/test.parquet","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"data/test/test.parquet","truncatedAt":512}],"sampleObjects":["NSynth.py","data/test/test.parquet","data/train/batch_0.parquet","data/train/batch_1.parquet","data/train/batch_10.parquet"],"sidecars":[{"bytes":"---\nlicense: cc-by-4.0\ndataset_info:\n  features:\n  - name: id\n    dtype: string\n  - name: note\n    dtype: int64\n  - name: note_str\n    dtype: string\n  - name: instrument\n    dtype: int64\n  - name: instrument_str\n    dtype: string\n  - name: pitch\n    dtype: int64\n  - name: velocity\n    dtype: int64\n  - name: sample_rate\n    dtype: int64\n  - name: qualities\n    sequence: int64\n  - name: qualities_str\n    sequence: string\n  - name: instrument_family\n    dtype: int64\n  - name: instrument_family_str\n    dtype: string\n  - name: instrument_source\n    dtype: int64\n  - name: instrument_source_str\n    dtype: string\n  - name: audio\n    dtype: audio\n  splits:\n  - name: train\n    num_bytes: 129511245\n    num_examples: 289205\n  - name: validation\n    num_bytes: 5679042\n    num_examples: 12678\n  - name: test\n    num_bytes: 1830670\n    num_examples: 4096\n  download_size: 25233566634\n  dataset_size: 137020957\ntask_categories:\n- audio-to-audio\n- audio-classification\ntags:\n- music\npretty_name: NSynth\nsize_categories:\n- 100K<n<1M\n---\n# Dataset Card for NSynth\n\n<!-- Provide a quick summary of the dataset. -->\n\nThe NSynth dataset is an audio dataset containing over 300,000 musical notes across over 1000 commercially-sampled instruments, distinguished by pitch, timbre, and envelope. Each recording was made by playing and holding a musical note for three seconds and letting it decay for one second. The collection of four-second recordings ranges over every pitch on a standard MIDI piano (or as many as possible for the given instrument), played at five different velocities.\nThis dataset was created as an attempt to establish a high-quality entry point into audio machine learning, in response to the surge of breakthroughs in generative modeling of images due to the abundance of approachable image datasets (MNIST, CIFAR, ImageNet). NSynth is meant to be both a benchmark for audio ML and a foundation to be expanded on with future datasets.\n\n### Dataset Description\n\n<!-- Provide a longer summary of what this dataset is. -->\n\nSince some instruments are not capable of producing all 88 pitches in the MIDI piano's range, there is an average of 65.4 pitches per instrument. Furthermore, the commercial sample packs occasionally contain duplicate sounds across multiple velocities, leaving an average of 4.75 unique velocities per pitch.\n\nEach of the notes is annotated with three additional pieces of information based on a combination of human evaluation and heuristic algorithms:\n  1. Source: The method of sound production for the note’s instrument. This can be one of `acoustic` or `electronic` for instruments that were recorded from acoustic or electronic instruments, respectively, or `synthetic` for synthesized instruments.\n|Index|ID|\n|:----|:----|\n|0|acoustic|\n|1|electronic|\n|2|synthetic|\n\n\n\n\n\n  2. Family: The high-level family of which the note’s instrument is a member. Each instrument is a member of exactly one family. See the complete list of families and their frequencies by source below.\n|**Index**|**ID**|\n|:---|:---|\n|0|bass|\n|1|brass|\n|2|flute|\n|3|guitar|\n|4|keyboard|\n|5|mallet|\n|6|organ|\n|7|reed|\n|8|string|\n|9|synth_lead|\n|10|vocal|\n\n|**Family**|**Acoustic**|**Electronic**|**Synthetic**|**Total**|\n|:----|:----|:----|:----|:----|\n|Bass|200|8387|60368|68955|\n|Brass|13760|70|0|13830|\n|Flute|6572|35|2816|9423|\n|Guitar|13343|16805|5275|35423|\n|Keyboard|8508|42645|3838|54991|\n|Mallet|27722|5581|1763|35066|\n|Organ|176|36401|0|36577|\n|Reed|14262|76|528|14866|\n|String|20510|84|0|20594|\n|Synth Lead|0|0|5501|5501|\n|Vocal|3925|140|6688|10753|\n|**Total**|108978|110224|86777|305979|\n\n  3. Qualities: Sonic qualities of the note. See below for descriptions of the qualities, and [here](https://magenta.tensorflow.org/datasets/nsynth#quality-co-occurrences) for information on co-occurences between qualities.\n\n|**Index**|**ID**|**Description**|\n|:----|:----|:----|\n|0|`bright`|A large amount of high frequency content and strong upper harmonics.|\n|1|`dark`|A distinct lack of high frequency content, giving a muted and bassy sound. Also sometimes described as ‘Warm�","encoding":"utf8","key":"README.md","name":"README.md","size":10069,"truncated":true}],"sourceObjectKey":"data/test/test.parquet","starterCell":"from datasets import load_dataset\nds = load_dataset(\"jg583/NSynth\")\nprint(ds)","starterSamplePath":"data/test/test.parquet","totalObjects":40},"summary":"NSynth is an audio dataset of over 300,000 musical notes from 1,006 instruments from commercial sample libraries, released by Google Magenta to give audio machine learning a…","title":"NSynth: Neural Audio Synthesis musical notes dataset","updatedAt":"2026-11-14T17:40:49.534Z","visibility":"public","websiteUrl":"https://magenta.tensorflow.org/datasets/nsynth"}}