An Audio-Visual Framework for Transcription and Fingering Optimization in Bass Guitar Performance

Fingering plays a crucial role in musical performance, yet most computational approaches focus on piano, overlooking the spatial complexity of string instruments like the bass.This paper presents a framework for estimating optimal bass fingering by analyzing video input, detecting strings, frets, finger positions, and played notes.Using a Bi-LSTM model, the system estimates optimal fingering sequences.The framework supports both user-recorded and online videos, offering a scalable solution for bass performance analysis.