Level-dependent changes in perception of speech envelope cues

Abstract

Level-dependent changes in temporal envelope fluctuations in speech and related changes in speech recognition may reveal effects of basilar-membrane nonlinearities. As a result of compression in the basilar-membrane response, the "effective" magnitude of envelope fluctuations may be reduced as speech level increases from lower level (more linear) to mid-level (more compressive) regions. With further increases to a more linear region, speech envelope fluctuations may become more pronounced. To assess these effects, recognition of consonants and key words in sentences was measured as a function of speech level for younger adults with normal hearing. Consonant-vowel syllables and sentences were spectrally degraded using "noise vocoder" processing to maximize perceptual effects of changes to the speech envelope. Broadband noise at a fixed signal-to-noise ratio maintained constant audibility as speech level increased. Results revealed significant increases in scores and envelope-dependent feature transmission from 45 to 60 dB SPL and decreasing scores and feature transmission from 60 to 85 dB SPL. This quadratic pattern, with speech recognition maximized at mid levels and poorer at lower and higher levels, is consistent with a role of cochlear nonlinearities in perception of speech envelope cues.

FIG. 1

One-third-octave band spectra of vocoded CV syllables (blue lines), sentences (red lines), and background noise (thick black lines) for five overall speech levels (45, 50, 60, 70, and 85 dB SPL). Mean quiet thresholds are also shown in each panel (triangles).

FIG. 2

Mean (thick lines) and individual (thin lines) recognition scores plotted as a function of speech level, for consonants (blue, top) and key words in sentences (red, middle). Mean scores for the two speech materials are also displayed in the bottom panel. For clarity, some data points are offset along the abscissa. Error bars indicate ±1 standard deviation.

FIG. 3

Key word recognition scores plotted against consonant recognition scores, for five speech levels (top to bottom panels). Pearson correlation coefficients and linear regressions are included in each panel.

FIG. 4

Top: mean information transmitted plotted as a function of speech level for the three acoustic-phonetic features of voicing, manner of articulation, and place of articulation. Bottom: same as top panel, but for three sub-categories of manners of articulation (plosive, nasality, and frication).

FIG. 5

Slope (percent per dB) at the highest speech level (85 dB SPL) plotted against slope at the lowest speech level (45 dB SPL), for recognition of consonants (top) and key words in sentences (bottom). Slopes for each speech material were computed from the polynomial fit applied to the score-level function for each subject. Pearson correlation coefficients and linear regression functions are included in each panel.

FIG. 6

Slope (percent per dB) calculated from speech scores at the highest level (85 dB SPL) plotted against the range of scores for recognition of consonants (top) and key words in sentences (bottom). Pearson correlation coefficients and linear regression functions are included in each panel.

FIG. 7

Top: DPOAE levels plotted as a function of L ₂ for f ₂ of 1.0 kHz. DPOAE input–output function slopes were computed from DPOAE levels recorded for L ₂ between 40 and 65 dB SPL (red lines). Bottom: slopes of DPOAE input–output functions for f ₂ at 1.0 kHz (from the top panel) plotted against DPOAE summed levels. The Pearson correlation coefficient and linear regression function are also included.

FIG. 8

Slopes of DPOAE input–output functions for f ₂ of 2.0 kHz plotted against the range of key word recognition scores. The Pearson correlation coefficient and linear regression function are also included.

FIG. 9

Top: Slopes of DPOAE input–output functions for an f ₂ of 1.0 kHz plotted against the change in consonant recognition scores with speech level increasing from 60 to 70 dB SPL. Bottom: slopes of DPOAE input–output functions for an f ₂ of 2.0 kHz plotted against the change in key word recognition scores with speech level increasing from 60 to 70 dB SPL. Pearson correlation coefficients and linear regression functions are included in each panel.