Importance of temporal-envelope cues in consonant recognition

Area of Science:

• Auditory Neuroscience
• Speech Processing
• Psychoacoustics

Background:

• Understanding speech perception relies on analyzing temporal modulations in the speech envelope.
• The relative importance of different modulation frequency regions for consonant recognition remains an active area of research.

Purpose of the Study:

• To investigate the role of specific modulation frequencies (0-20 Hz) in the speech envelope for consonant recognition.
• To determine the impact of filtering distinct modulation frequency bands on the perception of phonetic features.

Main Methods:

• Vowel-consonant-vowel (VCV) syllables were manipulated by replacing fine structure with noise.
• Temporal envelopes were notch filtered at center frequencies of 8, 12, and 16 Hz with varying widths (4, 8, 12 Hz).
• Consonant recognition was tested in ten normal-hearing subjects, with results analyzed using recognition scores, multidimensional scaling (INDSCAL), and sequential information analysis (SINFA).

Main Results:

• Consonant recognition demonstrated robustness to the removal of certain modulation frequencies.
• Significant speech degradation occurred only when a wide notch filter (around 8 Hz) was applied.
• Voicing information was lost, while plosiveness and nasality perception were differentially affected by filtering.

Conclusions:

• Low-modulation frequencies, particularly around 8 Hz, are critical for overall speech intelligibility and the perception of specific consonant features.
• Filtering even narrow bands of low modulation frequencies can substantially impact the transfer of features like plosiveness and nasality.