Rhythm Guitar Tablature Continuation through Picking Pattern Generation - Data

This record contains the data that was used to train the models in the following paper: D'Hooge, A., Déguernel, K., and Bigo, L., "Rhythm Guitar Tablature Continuation through Data-Driven Picking Pattern Generation". Accepted for publication in the TISMIR journal. Because this data is extracted from the DadaGP Dataset that is available upon request for research purposes, we license our derived data under a CC BY-NC-SA 4.0 license (refer to the license file for more details) and also make it available upon request for non-commercial purposes. In this record you will find: DadaGP-RhythmGuitarAnalysis: a large CSV file (compressed to zip) containing the results of processing the entire DadaGP dataset with the approach described in Régnier et al. 2021 pretrain.zip: The preprocessed data splits used for the first training phase described in the paper (4-bar tab sequences without chord diagrams) refine_subset.zip: The preprocessed data splits used for the second training phase described in the paper (4-bar tab sequences with annotated chord diagrams) vocabulary_tokens.json: the dictionary that maps token integers to what they represent. The code and the data obtained from the subjective study are publicly released on Github