Every instrument has a unique sound, and every player has their own technique to shape that sound. The extraordinary happens when the two elements come together to fuse in an emotional high. The auditory experience of such a sonic entanglement can vary a lot, and this is particularly true for the guitarist. It is therefore practically impossible to collect enough data that is representative of all instruments, all players, all techniques, and all recording environments to train an algorithm that would be able to decipher the message hidden in the music, not to mention the resources such an amount of data would require for storage and processing. To nevertheless provide you with a practical solution, we at Algoriffix have approached the problem from a different angle: We have spent years of research to understand the psychoacoustic phenomena behind pitch perception and its physical correlates and have studied how machine learning algorithms “see” pitch. Based on our findings, we have broken down the problem into its basic elements and developed a tool that gives you the freedom to parametrize a model that works best for you and your instrument — no cloud, no big data, no copyright infringement. Good vibes only!
This tool is Transkr V2. It combines the domains of machine learning and digital signal processing to enable artificial musical intelligence. Besides polyphonic pitch recognition, the plug-in also features:
– Beat, key, chord and note tracking
– Support for multiple instruments, including voice
– Parametric model customization and fine-tuning
– A polyphonic tuner
– A fretboard with guitar fingerings
– Low latency
– Audio-to-MIDI file conversion
