Effects of Fundamental Frequency and Vocal Tract Resonance on Sentence Recognition in Noise

This study examined the effects of change in a talker's sex-related acoustic properties (fundamental frequency [F0] and vocal tract resonance [VTR]) on speech recognition in noise.

The stimuli were Hearing in Noise Test sentences, with the F0 and VTR of the original male talker manipulated into four conditions: low F0 and low VTR (L_F0L_VTR; i.e., the original recordings), low F0 and high VTR (L_F0H_VTR), high F0 and high VTR (H_F0H_VTR), and high F0 and low VTR (H_F0L_VTR). The listeners were 42 English-speaking, normal-hearing adults (21-31 years old). The sentences mixed with speech spectrum-shaped noise at various signal-to-noise ratios (i.e., -10, -5, 0, and +5 dB) were presented to the listeners for recognition.

The results revealed no significant differences between the H_F0H_VTR and L_F0L_VTR conditions in sentence recognition performance and the estimated speech reception thresholds (SRTs). However, in the H_F0L_VTR and L_F0H_VTR conditions, the recognition performance was reduced, and the listeners showed significantly higher SRTs relative to those in the H_F0H_VTR and L_F0L_VTR conditions.

These findings indicate that male and female voices with matched F0 and VTR (e.g., L_F0L_VTR and H_F0H_VTR) yield equivalent speech recognition in noise, whereas voices with mismatched F0 and VTR may reduce intelligibility in noisy environments.

https://doi.org/10.23641/asha.29052305.