Spectral Translation
Map image position and brightness into audible frequency bands. Inspect the generated sound as a measured spectrogram. For reversible pixel encoding, use Cipher and Reconstruction.
Try an image, adjust the settings and download your sound. Uploaded previews stay in this browser; library, generated and gallery sources use public links. Guest experiments are not saved. Sign in for saved studies, collections and publishing.
Each translation interprets a 256 × 256 image sample over 8 seconds. Download your sound before leaving.
How this mapping works
Image structure v1 analyzes a 256 × 256 sample. Bright regions feed low-to-high frequency bands; edge detail adds texture, repeated spacing shapes pulses, and vertical strokes make more articulated sound than horizontal layers. Harmonics sets band count, Fidelity emphasizes bright regions, Grain controls edge texture, and Encoding Density controls pulse speed. The same settings and pixels reproduce the same sound.
Drop image or click to select
PNG · JPG · WEBP · GIF · Local preview
Sound character (advanced)
Adds brighter, more complex tones.
Duration stretches the image scan and note spacing. Use Audio finishing for channel placement, output gain and loop edge fades.
Higher values raise the pitch range.
Higher values make the sound louder.
Higher values add more frequency bands.
Higher values increase band strength.
Higher values add stronger pulsing.
Higher values speed up pulsing when grain is active.
Audio finishing
Edge fading brings both ends to silence without changing duration. It softens the join but may create a rhythmic dip. MIDI and note diagrams describe the original sequence before audio finishing.