Introduction
Arvid Linde first used AI to clean up an old demo recording. The hiss disappeared, but so did something else: a room tone, a little dirt, the sense that the song had been played by tired people in a real place.
Arvid lives in Berlin and works with independent musicians on fair audio-AI workflows. He came through touring, studio improvisation and the worry that sound can be separated too easily from consent.
Story of the Path into AI
Audio restoration fascinated him because it could rescue material that would otherwise remain unusable. But the tools also invented textures that had never existed. Arvid had to distinguish technical improvement from artistic falsification.
He learned audio models, licensing issues and transparent production notes. His first project was a restored EP where every AI intervention was documented and compared with the original tracks. One track sounded cleaner and less truthful. He kept the noise.
Rights questions soon became central. Voice synthesis, samples and style imitation opened possibilities and dangers. Arvid refused one commission because consent from the singer was unclear. Instead he built an instrument from his own field recordings.
Current Work
Today Arvid helps musicians decide where AI belongs in their workflow: restoration, separation, arrangement sketches, accessibility, documentation. He insists that provenance and consent are part of the sound, not paperwork after the fact.
His work shows that AI in studios can be creative without treating recordings as free raw material. The best results often come when the machine’s output is compared against memory: what did that rehearsal feel like, what did the room do, what should remain imperfect?
Personal Advice
“Sound is memory. Do not treat it like free raw material,” Arvid says. He advises musicians to document AI use clearly and to protect the voices of people who are not in the room to object.
Key Facts
Age and place: 42, Berlin.
Background: independent music scene, rights questions, artistic integrity.
Entry into AI: restored EP with documented AI interventions.
Focus today: music and AI.
Typical tools: audio restoration, generative models, license documentation.
Werkstattnotiz
Arvid keeps two versions of the same track: one cleaned too far, one left with a stubborn hiss. The second moves him more. He is now testing how production notes can describe not only what AI changed, but what the artist chose to leave damaged.