Learning-Based Methods for Comparing Sequences, with Applications to Audio-to-MIDI Alignment and Matching

By Colin Raffel. 2016

353 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Signal Processing

Citations in the corpus

Neural Audio Synthesis of Musical Notes with WaveNet Autoencoders (2017, background): “Despite our best eﬀorts to convey analysis in plots, listening to the samples is essential to understanding this paper and we strongly encourage the reader to listen along as they read.”
A Hierarchical Latent Vector Model for Learning Long-Term Structure in Music (2018, methodology): “We ﬁrst evaluated the LMD-trained melody model on a subset of the full evaluation set made by excluding any examples in the LMD train set.”
MuseGAN: Symbolic-domain Music Generation and Accompaniment with Multi-track Sequential Generative Adversarial Networks (2017, methodology): “We employed the matched subset of the Lakh MIDI dataset (LMD)1 (Raffel 2016) as the training dataset, for it provides a rich collection of real-world MIDI files and some associated meta-data.”
LakhNES: Improving Multi-instrumental Music Generation with Cross-domain Pre-training (2019, background): “In contrast, the Lakh MIDI dataset [4] is enormous ( 175 k songs) but heterogeneous (varying numbers of instruments per piece).”
A Continuous Time Framework for Discrete Denoising Models (2022, methodology): “3 Monophonic Music We generate our training dataset from the Lakh pianoroll dataset [28, 29] (license CC By 4.”
MuseGAN: Multi-track Sequential Generative Adversarial Networks for Symbolic Music Generation and Accompaniment (2017, methodology): “The piano-roll dataset we use in this work is derived from the Lakh MIDI dataset (LMD) (Raffel 2016),4 a large collection of 176,581 unique MIDI files.”
Invariances and Data Augmentation for Supervised Music Transcription (2017, methodology): “This alignment can be constructed using an optimal-alignment protocol, as in the SyncRWC [4], Lakh [6], and MusicNet [7] datasets.”
Temporal pattern attention for multivariate time series forecasting (2018, result): “LPD-5-Cleansed, however, was not split in previous work (Dong et al. 2018; Raffel 2016); thus we randomly split it into training (80%), validation (10%), and testing (10%) sets.”
Learning a Latent Space of Multitrack Measures (2018, methodology): “Our models are all trained on the Lakh MIDI Dataset [30], a collection of 176,581 MIDI files scraped from the web.”