AG-PT-set: Acoustic Guitar Playing Technique dataset

This is the Acoustic Guitar Playing Technique dataset (AG-PT-set).It contains 15 hours and 55 minutes of monophonic recordings of 12 expressive guitar playing techniques (pitched and percussive).  Of these, 10 hours and 4 minutes of recordings encompassing 8 of the 12 techniques have been labeled, meaning that onsets were identified at the millisecond level and their timestamp was saved, alongside playing technique information.As a result, 32,592 individual notes have been labeled.Recordings were performed by 6 players on 7 different acoustic steel-string guitars.

The techniques are the following:

Kick technique (*percussive*): producing a sound that resembles a kick drum by hitting the lower right part of the top of the guitar body;

Snare-A technique (*percussive*): producing a sound by hitting the lower right side of the guitar body;

Tom technique (*percussive*): producing a sound by hitting the area of the guitar body near the top of the end of the fretboard, using the thumb;

Snare-B technique (*percussive*): producing a sound by hitting the muted strings over the end of the fretboard; 

Natural Harmonics (pitched): plucking the strings while lightly touching the string with the fretting finger (i.e., not pressing the string fully), therefore letting only some harmonic overtones ring;

Palm Mute: partially muting the strings with the palm of the picking hand, resulting in a muffled sound.

Pick Near Bridge (pitched): plucking the string near the guitar bridge, producing sounds with great high-frequency content;

Pick Over the Soundhole (pitched): plucking the string over the soundhole, producing sounds with lower treble content and greater intensity;

Bending technique (pitched): pulling the strings, raising the pitch (half-tone interval);

Hammer-on technique (pitched): sharply bringing a finger down onto the fingerboard, creating a legato sound (half-tone interval);

Staccato (pitched): playing short notes;

Vibrato (pitched): Moving the fretting finger to warp the pitch and tone of the sound.

Techniques 1 through 8 have been meticulously labeled.

This dataset is presented in detail in the following conference paper:

D. Stefani, G. A. Giudici, and L. Turchet. 2024. On the Importance of Temporally Precise Onset Annotations for Real-Time Music Information Retrieval: Findings from the AG-PT-set Dataset. In Proceedings of the 19th International Audio Mostly Conference: Explorations in Sonic Cultures (AM '24). 

https://doi.org/10.1145/3678299.3678325