Showcasing modern audio-to-MIDI technology at GotGuitars 2026.
What happens when a guitar performance stops being just sound and becomes editable musical structure?
That question became the focus of a small live workshop hosted by Algoriffix at Gothenburg’s largest guitar fair, GotGuitars.
The setup was deliberately simple: a guitar, a microphone, a DAW, two virtual instruments, and two audio-to-MIDI plug-ins — Transkr V2 for polyphonic guitar input, and Pitch Perfekt for monophonic voice.
But the workshop was not really about plug-ins.
It was about what happens when sound becomes MIDI data.
Beneath the Waveform
Most music software operates directly on audio waveforms.
You can apply effects like pitch shifting, equalisation, compression, distortion, or spatial processing. But the underlying composition usually remains engraved into the waveform.
Audio-to-MIDI shifts manipulation away from the waveform and into the musical structure underneath.
Once note events and articulation have been extracted from an acoustic performance, the music becomes editable independently of the original recording. Notes can be rearranged, harmonies layered, and synth patches switched without rerecording the original material.
This shift — from manipulating sound to manipulating musical structure — became one of the central themes of the workshop.
From Guitar Recordings to Musical Arrangement
To demonstrate the studio workflow, several guitar layers were recorded beforehand together with jazz guitarist Luka Elez:
- a steady bass line,
- syncopated bossa nova chords,
- sustained jazz harmonies,
- and rhythmic comping with mutes.
The recordings were analysed with Transkr V2, converted into MIDI data, and imported into the DAW.
Once represented as MIDI, the arrangement could be reshaped freely. Timing could be refined, notes adjusted, and instruments switched without touching the original recordings.
The resulting 16 bars became the foundation for the live demonstration.
Interpreting Fingerstyle in Real-Time
The live demonstration began with playback of the reconstructed 16-bar arrangement created from Luka Elez’s guitar recordings.
Different synth patches were switched live during playback, allowing the audience to hear how the same musical material could drive entirely different textures and timbres with latency below 20 milliseconds.
Midway through the arrangement, a sudden chord slide emerged from the backing track — a gesture created entirely at the MIDI level and difficult to reproduce through conventional audio processing alone.
The reaction in the room was immediate.
Both the audience and Luka paused for a split second before smiling and asking:
“What just happened?”
That moment captured the central idea behind the workshop more clearly than any technical explanation could. Once sound becomes MIDI data, musical structure becomes something you can directly reshape.
Only afterwards did Luka begin improvising live over the looping arrangement while the incoming DI signal was translated into MIDI events and routed through virtual instruments inside the DAW.
The demonstration was not about autonomous AI music generation. The system was not composing on its own. The focus was on translating acoustic performance into playable notes quickly enough to remain usable in live performance.
That distinction became the focal point of the discussion that followed.
Many of the questions from the audience focused less on artificial intelligence and more on tracking accuracy, latency, DAW integration, signal quality, workflow compatibility, and practical usability on stage.
The sound technician responsible for the demo room mentioned that he had experimented with audio-to-MIDI technology years earlier but never found it convincing enough to use in production. After the demonstration, however, he admitted that the naturalness of the backing track made him want to revisit the technology again.
Featuring jazz guitarist Luka Elez
Instagram: @luka__elez
Voice-to-MIDI and Harmonisation
The workshop also included a short demonstration of Pitch Perfekt.
Simple vocal lines were sung into the microphone while the detected pitch was used to generate harmonised chords automatically.
Together with the guitar demonstration, this highlighted another important aspect of music analysis: once musical structure has been extracted from sound, the original sound source no longer determines the sonic identity of the result.
A vocal phrase can control layered synthesizers. A guitar can drive orchestral textures. The same musical gestures can shift between entirely different sonic identities through changes in synth patches or virtual instruments.
A guitarist can also switch between entirely different synthesizer setups during performance using a MIDI pedalboard while playing the same notes.
Musical Control Through MIDI
One of the more interesting aspects of the workshop was how quickly the discussion shifted away from software and toward musical control.
Once sound becomes MIDI data, entirely different workflows become possible:
- arrangement,
- instrumentation,
- and live instrument control.
Ultimately, the workshop was not really about software.
It was about what becomes possible once performance stops being locked inside the waveform.
Downloads
Audio and video material embedded in this article may not be reused or redistributed without prior written permission from the authors.
Leave a Reply