Model licences — Find the BPM of a song (bpm-finder) ==================================================== 1. Beat This! — beat and downbeat tracker (checkpoint "final0") ---------------------------------------------------------------- Authors: Francesco Foscarin, Jan Schlüter, Gerhard Widmer Organisation: Institute of Computational Perception, Johannes Kepler University (JKU) Linz, Austria Source: https://github.com/CPJKU/beat_this Licence: MIT. The upstream README states: "The code and the published model weights are released under the MIT license." (read 2026-09-25) Paper: Francesco Foscarin, Jan Schlüter, Gerhard Widmer. "Beat this! Accurate beat tracking without DBN postprocessing." Proceedings of the 25th International Society for Music Information Retrieval Conference (ISMIR 2024), San Francisco, CA, USA, 2024. Training data: according to the upstream README, "final0" was trained on 16 beat-tracking datasets (all except GTZAN). The terms of those datasets were not reviewed individually; this page relies only on the licence of the published weights. Full licence text (https://github.com/CPJKU/beat_this/blob/main/LICENSE): MIT License Copyright (c) 2024 Institute of Computational Perception, JKU Linz, Austria Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. 2. ONNX export used by this page: musetric/beat-this-onnx ---------------------------------------------------------- Repository: https://huggingface.co/musetric/beat-this-onnx (HF licence tag: mit) Files used: beat_this.onnx 120,259,561 bytes sha256 d6b41a44dbf555e90593f60dc86aea3689e1f5db427956e4c9036c8dfde970e8 mel-filterbank.bin 262,656 bytes sha256 1ee975d96f44ccf2c3bfe37825c1c1f0b089f5703c7a12a84b1f0a3bce004533 Export notes (from the model card): ONNX export of the JKU Beat This! checkpoint "final0", float32, opset 17. The graph is a static 513-frame rewrite (convolutions as Pad + im2col Gather + MatMul, Split lowered to Slice, stem BatchNorm folded) made to avoid silent WebGPU errors on some mobile GPUs; it takes exactly one [1, 513, 128] log-mel window per call. mel-filterbank.bin is torchaudio's slaney mel filterbank [513, 128] (f_min 30 Hz, f_max 11 kHz) written verbatim. The card states: "This export preserves the upstream MIT license; we do not claim authorship of the original weights." Feature extraction, windowing (chunk 513, border 6, keep_first), peak picking, tempo and meter are implemented by this page (engine.js / assets/rhythm.js), following the reference implementation. 3. Runtime libraries (loaded from this site, not from third parties) -------------------------------------------------------------------- ONNX Runtime Web — Copyright (c) Microsoft Corporation — MIT License https://github.com/microsoft/onnxruntime/blob/main/LICENSE Transformers.js — Copyright (c) Hugging Face — Apache License 2.0 (loaded by the shared worker) https://github.com/huggingface/transformers.js/blob/main/LICENSE 4. Sample audio --------------- samples/drum-loop-120-4-4.mp3 and samples/waltz-150-3-4.mp3 were synthesised for this page from sine waves and noise (tools/bpm-finder/probe/make_samples.py) and are dedicated to the public domain under CC0 1.0 (https://creativecommons.org/publicdomain/zero/1.0/).