Audio Analysis
aerial analyzes every sample in your library across 8 distinct audio attributes. These attributes drive the scatter plot positioning and let you search for samples by their sonic characteristics rather than by filename.
The 8 Attributes
| Attribute | Measures | Low Values | High Values |
|---|---|---|---|
| Pitch | Fundamental frequency (Hz) | Low/bass sounds | High/treble sounds |
| Brightness | Spectral centroid | Dark, warm sounds | Bright, present sounds |
| Length | Effective duration (seconds) | Short hits | Long sustains |
| Attack | Onset to peak time | Percussive, snappy | Slow swells, pads |
| Inharmonicity | Overtone deviation | Pure, harmonic tones | Bells, metallic, noisy |
| Richness | Spectral spread | Thin, simple sounds | Rich, complex sounds |
| Spectral Flux | Rate of timbral change | Static, steady tones | Evolving textures |
| Sustain Level | Steady-state RMS vs. peak | Quick decay sounds | Sustained, held notes |
Attribute Details
Pitch
The fundamental frequency of the sample, detected via autocorrelation. Displayed in Hz. This is useful for sorting samples by musical note or finding samples in a specific pitch range. Samples without a clear tonal center (noise, complex textures) may show less reliable pitch values.
Brightness
The spectral centroid, which represents the center of mass of the frequency spectrum. A higher spectral centroid means more energy in the upper frequencies, producing a brighter, more forward sound. Lower values indicate darker, warmer timbres.
Length
The effective duration of the sample in seconds, measured from the start to the point where energy drops below a threshold. This is more meaningful than raw file length because it ignores trailing silence.
Attack
The time from the onset of the sound to its peak amplitude. A short attack time means a percussive, immediate sound (clicks, plucks, drums). A long attack time indicates a gradual swell (pads, strings, risers).
Inharmonicity
Measures how much the overtones of a sound deviate from perfect harmonic ratios. A value of 0 indicates a perfectly harmonic tone (like a pure sine wave or clean string). Higher values indicate sounds with stretched or misaligned partials, such as bells, metallic percussion, or inharmonic synthesis.
Richness
The spectral spread or bandwidth of the sound. Narrow spectral spread produces a thin, simple tone with energy concentrated in a small frequency range. Wide spectral spread means a rich, complex sound with energy distributed across the spectrum.
Spectral Flux
The rate of timbral change over the duration of the sample. High spectral flux means the frequency content is constantly shifting - typical of evolving textures, vocal samples, or modulated sounds. Low spectral flux indicates a static, unchanging tone.
Sustain Level
The steady-state RMS energy of the sample relative to its peak amplitude. A high sustain level means the sample maintains its energy over time (organ, bowed strings). A low sustain level means the sound decays quickly after its initial transient (plucks, percussion).
How Analysis Works
Analysis runs automatically whenever you add a new sample folder. aerial processes each audio file in the background, extracting all 8 attributes in a single pass.
Results are cached locally, so re-scanning the same folder is instant. The cache persists between sessions, and you only pay the analysis cost once per sample file.