// GHOST FREQUENCIES — recursive listening //
For Track 05. ⠠⠤⠤⠤ (Ghost Frequencies) I wanted to try something that bends authorship, hearing, and interpretation into a single workflow — a kind of algorithmic tinnitus, where the track is built from layers I can’t reliably hear, but still shape.
This one starts, as usual, with the Zoom H2n: air recordings, induction hums, contact-mic subtones. Those become the “gravity seeds” fed into SunoAI, alongside prompts shaped around phantom tones, misphased harmonics, and Hildur-style sinestacks.
After several prompt tweaks and resets (to stop Suno drifting into melody or smoothing out the pressure-field), I downloaded the v1.0 Suno mix and ran it through the Between Lines app — sound ➝ SRT ➝ sound — using the Ritual Dread preset (Phrygian, low-mid band, long sustains, Beksínski energy, dark chapel resonance).
From there, the process becomes recursive:
1 → Original Suno track 2 → Convert to SRT 3 → Convert the SRT back into audio 4 → Convert that into a new SRT 5 → Generate another audio layer 6 → Repeat until the ghosts accumulate
Each pass becomes less like “sound” and more like the memory of sound. The text captions dissolve into drones. Rhythm collapses into pressure. The recursion starts inventing its own internal logic — meaning without signal.
Then into Adobe Audition, where these three BL passes sit under the main track like spectral afterimages. Except here’s the complication: my hearing devices auto-adjust. CI compression boosts low-level detail, kills dynamic range, flattens tonal edges, and exaggerates high-mids. I can’t trust what I’m hearing. What’s “quiet” to me may be overpowering to someone else. What feels balanced may actually be skewed.
So mixing becomes a collaboration with the machine:
• High-pass all BL layers above 800 Hz • Low-pass them below 4.5 kHz • Reduce gain (–28 dB, –30 dB, –33 dB) • Trust the meters more than my ears • Trust the process more than perception
The recursive layers end up inaudible to me, but they’re there — shaping the air of the track, adding phantom motion, spectral grit, pressure swells, and the kind of tinnitus-adjacent shimmer that lives “just outside” hearing.
Ghost Frequencies becomes less about composing sound and more about constructing something I can’t fully perceive — a piece that sits in the gap between how I hear (aided, compressed, adaptive) and how others might hear (full dynamic fidelity).
Deaf/apd workflow as process. Machine-hearing as collaborator. Signal becoming memory becoming noise.
This is Signal // Noise in practice.
