Guitar Tablature Transcription with Deep Transformer Models, for Igor Sok

Automatic guitar tablature transcription is an active field in music information retrieval (MIR). It entails extracting guitar-specific music annotations from pieces of audio recordings of guitar music. Compared to other instruments such as the piano, this field is relatively underdeveloped. This is mainly due to the lack of large, high-quality datasets.
Several approaches have come forward to combat this issue, but the problem remains underexplored. The main challenges this project aims to tackle are the lack of data and the exploration of transformer models utilised for automatic tablature transcription. This entails exploring brand-new datasets such as GAPS and addressing the overfitting to the GuitarSet dataset that is very prevalent in the field, as it is one of the only datasets with a sizeable amount of richly annotated guitar music recordings. Deep transformer models will be employed to transcribe pieces of guitar music. To do this, synthetic data will have to generated, as transformer models require a lot of training examples to be highly useful.