Humans detect spectral tilts as small as 1 dB across the audible range

Evidence

The amplitude vs. frequency curve, which is known as the frequency response, is the single most important measurable parameter of any audio device, and the general objective is a smooth flat curve extending over the audible frequency range. … There is evidence that we can detect slopes of about 0.1 to 0.2 dB/octave, which translates into a 1 to 2 dB tilt from 20 Hz to 20 kHz (Kommamura and Mori, 1983). Such a spectral error, if small, is likely to be quite benign and subject to adaptation: we simply would get used to it. — , Toole, Floyd E - Sound reproduction-…, 374

Categorization

• XREF: connects to: Trained listeners perceive flat spectra from loudspeakers whose on-axis response is flat; Masking noise prevents listeners from using unintended spectral cues

Related

• Trained listeners perceive flat spectra from loudspeakers whose on-axis response is flat (similarity 0.86)
• Listeners benefit from but do not fully use broadband spectral information. (similarity 0.83)
• Humans hear spectral content, not waveform shape or phase shifts (similarity 0.82)
• The auditory system largely ignores signal phase, driving spectral analysis conventions (similarity 0.81)
• Blind listening shows listeners consistently prefer flat, smooth anechoic frequency response (similarity 0.81)
• Perceived spectral balance matches on-axis response more than sound power or room curve (similarity 0.81)
• Simultaneous-masking filters may understate the ear’s true spectral resolution (similarity 0.81)
• Slight response deviations go unnoticed without an audible reference (similarity 0.81)
• Trained listeners can draw accurate spectral-balance curves of what they hear (similarity 0.80)
• Human hearing requires far more SPL to detect low-frequency tones than mid-range ones (similarity 0.80)
• Narrow spectral bands with steep slopes produce stronger pitch than narrow spectral gaps do (similarity 0.80)
• Added masking noise prevents off-frequency cues in hearing experiments (similarity 0.80)
• Off-frequency listening can distort auditory filter measurement unless masked (similarity 0.80)
• Listeners prefer flat amplitude response but are insensitive to phase shift (similarity 0.80)
• Real loudspeakers show flat magnitude response only below 800 Hz with a midrange dip (similarity 0.80)

Broader Connections

• Trained listeners perceive flat spectra from loudspeakers whose on-axis response is flat #corroborates (0.85)
• Masking noise prevents listeners from using unintended spectral cues #qualifies (0.80)