Internal noise, not feature extraction, explains individual variation in model vowel identification

Evidence

The model estimates relevant features from electrodograms generated by the CI sound processor. Because we are modelling a vowel identification task it makes sense to extract features related to formant frequencies. Moreover, because we are analyzing the effect of enhancing spectral contrast it seems logical to use spectral contrast features between formants. In this study the number of features was limited to two formants and therefore the MPI model is two dimensional. Next, the MPI model adds noise to the features. The variance of the noise is set based on the individual abilities of a CI user to perceive the features extracted from the electrodogram. In our implementation we used the results of the SMT task to set the noise variance. The obtained jnd from the SMT task was scaled between 0.001 and 0.5. This number was applied as the variance of the Gaussian noise applied in the MPI model and for this reason is termed jnd noise in Fig. 1 . — Dijk, Pim van, Physiology, Psychoacoustics and…, 1831 (#corroborates, via linked claim)

Categorization

• corroborated: 2 independent passages support this claim
• XREF: connects to: Vowel identification models inject noise scaled to each user’s spectral modulation threshold — Dijk, Pim van, Physiology, Psychoacoustics and…, 1831; Spectral modulation sensitivity at 0.5 cycles per octave best predicts vowel identification; Multi-dimensional phoneme identification models add noise scaled to individual perceptual acuity (+3 more)

Related

• Vowel identification models inject noise scaled to each user’s spectral modulation threshold (similarity 0.94)
• Multi-dimensional phoneme identification models add noise scaled to individual perceptual acuity (similarity 0.92)
• Just-noticeable-difference noise, not map settings, drives vowel-identification variability between cochlear implant users (similarity 0.87)
• MPI vowel-identification models add Gaussian jnd noise scaled from SMT thresholds (similarity 0.87)
• Increasing spectral contrast enhancement worsens vowel identification in CI users (similarity 0.81)
• Spectral modulation threshold declines track vowel identification declines (similarity 0.78)
• Over-attenuating spectral valleys may remove speech information CI users need (similarity 0.78)

Broader Connections

• Multi-dimensional phoneme identification models add noise scaled to individual perceptual acuity #corroborates (0.90)
• Vowel identification models inject noise scaled to each user’s spectral modulation threshold #corroborates (0.85) — Dijk, Pim van, Physiology, Psychoacoustics and…, 1831
• Spectral modulation sensitivity at 0.5 cycles per octave best predicts vowel identification #qualifies (0.75)
• Different noise types mask different specific speech sounds by matching their spectra #complicates (0.75)
• Classical spectral-peak vowel coding is especially vulnerable to noise and level changes #complicates (0.72)
• Sound-source differences dominate consonant perception variability more than listener uncertainty #complicates (0.70) — Dijk, Pim van, Physiology, Psychoacoustics and…, 5579