Open marja-w opened 4 months ago
Symbolic music generation:
Museformer: Transformer with Fine- and Coarse-Grained Attention for Music Generation: proposes a finer grained attention to capture the relation of bars in respect of the music structure, models longer sequences better, Code: https://github.com/microsoft/muzic
A Domain-Knowledge-Inspired Music Embedding Space and a Novel Attention Mechanism for Symbolic Music Modeling: propose fundamental music embedding (FME), embedding both, absolute and relative attributes, novel attention mechanism, repository: github.com/guozixunnicolas/FundamentalMusicEmbedding (not documented yet)
Pop music transformer: Beat-based modeling and generation of expressive pop piano compositions: used by fme as baseline
Mmm: Exploring conditional multi-track music generation with the transformer: this is basically what we want to do, but in midi format, might be interesting for the individualization idea, because it focuses on generating tracks separately, which allows to change them up independently, demo (no code available): https://jeffreyjohnens.github.io/MMM/
Compound word transformer: Learning to compose full-song music over dynamic directed hypergraphs
TunesFormer: Forming Tunes with Control Codes
language-based
Spotfys audio to midi converter: https://github.com/spotify/basic-pitch
Symbolic music generation:
Museformer: Transformer with Fine- and Coarse-Grained Attention for Music Generation: proposes a finer grained attention to capture the relation of bars in respect of the music structure, models longer sequences better, Code: https://github.com/microsoft/muzic
A Domain-Knowledge-Inspired Music Embedding Space and a Novel Attention Mechanism for Symbolic Music Modeling: propose fundamental music embedding (FME), embedding both, absolute and relative attributes, novel attention mechanism, repository: github.com/guozixunnicolas/FundamentalMusicEmbedding (not documented yet)
Pop music transformer: Beat-based modeling and generation of expressive pop piano compositions: used by fme as baseline
Mmm: Exploring conditional multi-track music generation with the transformer: this is basically what we want to do, but in midi format, might be interesting for the individualization idea, because it focuses on generating tracks separately, which allows to change them up independently, demo (no code available): https://jeffreyjohnens.github.io/MMM/
Compound word transformer: Learning to compose full-song music over dynamic directed hypergraphs
TunesFormer: Forming Tunes with Control Codes
language-based
Spotfys audio to midi converter: https://github.com/spotify/basic-pitch