Guitar TECHS
This dataset is designed for research in deep learning. This resource provides a vast diversity of techniques, musical fragments, chords, and scales for electric guitar, performed by different musicians in various recording environments. Its distinctive feature is its multimodal and multi-perspective content, which includes recordings from egocentric and exocentric microphones, alongside direct inputs and amplifier outputs. All data is synchronized with MIDI labels, allowing for the training of more robust models for specialized tasks such as automatic tablature transcription.
Freebok
This is a digital resource that functions as an open-access, centralized library, dedicated to providing the chord progressions for numerous jazz standards. Its purpose is to offer this reference material free of charge. The resource features a comprehensive song index, making it an invaluable tool for musicians, students, and arrangers focused on the study, improvisation practice, and analysis of the fundamental jazz repertoire.
Robust Guitar Tabs
This study is focused on Guitar Tablature Transcription (GTT). The research addresses the main limitation of this task, which is the lack of model robustness due to small datasets. To resolve this deficiency, the study proposes generating large-scale synthetic training data. Unlike previous approaches, this data is created from recordings of real guitar notes processed with various audio effects. The work demonstrates a significant improvement in GTT robustness by evaluating a model trained with this method on a new evaluation dataset composed of professional solo guitar performances.
Online Postonal Guitar
This website hosts three auxiliary web applications designed for the study and composition of post-tonal musical elements applied to the guitar. The platform includes a Chord Diagram Calculator for visualizing fingerings, an Accompaniment Generator that offers optimized voice leading, and an Interactive Prime Forms Visualizer in a three-dimensional space. These resources are valuable for guitarists looking to explore new harmonic possibilities and improve their theoretical understanding, proving useful in both tonal and non-tonal contexts.
EGFXSet
This dataset is composed of 8,970 recordings of electric guitar notes. It was obtained by enhancing the clean tones of a Fender Stratocaster with 12 popular analog effects, such as distortion, delay, and reverb, using real analog effects hardware. The set includes all possible notes in standard tuning and is exhaustively annotated with guitar parameters (string, fret, pickup configuration) and the exact settings of the effects used. Its purpose is to provide relevant data for musical information retrieval, offering an authentic and natural sound that overcomes the limitations of datasets based on digital emulations or software.