HertzKit MIDI Groove Lab — third-party notices Magenta.js 1.23.1 (Google Inc. and contributors), Apache License 2.0. TensorFlow.js 2.7.0 (Google LLC and contributors), Apache License 2.0. Pretrained model: groovae_2bar_humanize, downloaded unmodified from the official Magenta checkpoint hosting on 2026-10-08. https://github.com/magenta/magenta-js/blob/master/music/checkpoints/README.md https://storage.googleapis.com/magentadata/js/checkpoints/music_vae/groovae_2bar_humanize/ Official checkpoint documentation explicitly permits local use or self-hosting. https://magenta.tensorflow.org/groovae Groove MIDI Dataset: Google LLC, CC BY 4.0. No dataset audio/MIDI recordings are redistributed. Gillick, Roberts, Engel, Eck, Bamman. Learning to Groove with Inverse Sequence Transformations. ICML 2019. Dependencies are loaded without modification. HertzKit worker defines the regeneratorRuntime global before import to avoid the polyfill's dynamic evaluation fallback. No unsafe-eval is added to the site's CSP. HertzKit examples and drum synthesis are original code, without commercial samples. Bundled component notices: protobufjs, core-js, regenerator-runtime, long.js, seedrandom; see accompanying license files and license comments in tf.min.js. No affiliation or endorsement by Google/Magenta is implied. Downloaded source URLs: https://cdn.jsdelivr.net/npm/@magenta/music@1.23.1/es6/music_vae.js https://cdn.jsdelivr.net/npm/@tensorflow/tfjs@2.7.0/dist/tf.min.js SHA256 of unmodified distributed binaries and model files: 64096919d9452afc6404c441904f95c4b91f4da7e514f0725e858586fca94bc5 model/config.json 1612d3ed51d794229a33d5cc0cd563cf6aedb600d47d68c458147e215d499e3d model/group1-shard1of4 080df5f4cef89f2b17147f744e8e2c75b61c71a464094634ac87a20933696f2b model/group1-shard2of4 f323d2dec213af7693cd16f94bfe16e5028b6f9a24fc42db2b97baaedcae846f model/group1-shard3of4 b99d2e4ce5725bcb895a3699b25e4326a59ebf0a85e09cae3b78e3a8779463df model/group1-shard4of4 c124d59c9c8338c7722998a8f15301d82ffcce2aeaab3470f983f8f35152674a model/weights_manifest.json eb954854bebc70c0245552a2fe90c23a847a5a7b5bdd6ff38d97291598a8470b music_vae.js f04b77578a276dbcdcd8921c48a61a6090a3ada9de54d8cd848691a608f287d3 tf.min.js