A 65 dB sound can be "pleasantly quiet" or "unbearably squealing" – same level, completely different perception. Classical level measurement misses this. Psychoacoustics catches it.
Why dB does not suffice
dB(A) is frequency-weighted and good for occupational safety – but ignores masking, spectral distribution and temporal modulation. These determine whether a product sounds "premium" or "cheap".
The four key metrics
Loudness (Zwicker, ISO 532-1): perceived loudness in sone. 1 sone = 1 kHz tone at 40 dB.
Sharpness (acum): share of high-frequency energy. Above 2 acum sounds "sharp" – typical for e-drive whining.
Roughness (asper): fast amplitude modulation (15–300 Hz). Causes rattling/buzzing sensation.
Fluctuation strength (vacil): slow modulation (≤ 20 Hz), important for fan throbbing.
What the model sees that we do not immediately hear
Two pumps with identical dB(A): A produces 14 sone, B only 9 sone. B sounds quieter because its energy lies in a Bark band that is masked by its neighbour. Pure level measurement would not reveal this.
QA application
- Seat adjusters: loudness < 8 sone, sharpness < 1.8 acum
- Fans: fluctuation strength < 0.5 vacil
- E-drive whining: sharpness as additional criterion
Real-time computation in SonicTC
Zwicker models are compute-intensive (~20 ms per second classically). SonicTC uses an FPGA-accelerated implementation delivering all four metrics with < 2 ms latency.
Standards
| Standard | Metric | Status |
|---|---|---|
| DIN 45631 | loudness (Zwicker) | DE classic |
| ISO 532-1 | loudness (Zwicker) | international |
| ISO 532-2 | loudness (Moore-Glasberg) | auditory-near alternative |
| DIN 45692 | sharpness | standard |
| ECMA-74 | acoustic devices | IT industry |
