Speech-in-noise clarity thresholds worsen at positive reference signal-to-noise ratios

Evidence

The results are shown in Table 1 . The SNR JNDs at the three SNR references of ‑6, 0 and + 6 dB SNR were 2.8, 2.9, and 3.7 dB respectively; the first two were statistically no different and both less than the third [t (91.3) = 4.07; p = 0.0001 and t (121.4) = 4.11; p = 0.00007, respectively]. — Dijk, Pim van, Physiology, Psychoacoustics and…, 5676

Categorization

Category: Factual Claims

• QUESTION: Why does the SNR JND grow from ~2.8-2.9 dB at ≤0 dB reference to 3.7 dB at +6 dB? Possible ceiling effect, or reduced temporal fine-structure cues at high SNR.

Related

• SNR discrimination worsens at favorable signal-to-noise ratios (similarity 0.89)
• SNR just-noticeable difference grows from about 3 to 3.7 dB at positive reference SNRs (similarity 0.89)
• The just noticeable difference in signal-to-noise ratio for speech is about 3 dB (similarity 0.86)
• Normal-hearing listeners detect a 3 dB change in speech-to-noise ratio (similarity 0.84)
• Detection threshold for SNR change in sentences averages about 3 dB (similarity 0.84)
• Meaningful speech-in-noise benefit requires double the detectable SNR change (similarity 0.83)
• A higher required signal-to-noise ratio indicates worse speech-in-noise performance. (similarity 0.83)
• Effective SNR range for speech extends far beyond SII standard bounds. (similarity 0.83)
• Listeners value loudness increases as much as SNR improvements they provide (similarity 0.82)
• Speech needs at least 7.5 dB signal-to-noise ratio for adequate intelligibility (similarity 0.82)
• A just-noticeable difference in signal-to-noise ratio does not guarantee subjective value (similarity 0.82)
• Speech intelligibility saturates beyond certain signal-to-noise ratio bounds. (similarity 0.82)
• Speech-in-noise clarity discrimination can be measured with identical masked sentences while cueing levels (similarity 0.82)
• A 1 dB SNR improvement raises speech intelligibility by 7 to 19 percent. (similarity 0.81)
• Speech models incorrectly assume uniform SNR importance within fixed floor and ceiling bounds. (similarity 0.81)