Effects of fundamental frequency and vocal tract resonance on speech recognition in noise by non-native listeners

The present study examined the influence of changes in speakers’ fundamental frequency (fo) and vocal tract resonance (VTR) on speech recognition in different types of noise by non-native listeners. The goal was to identify whether the fo-VTR relationship has a similar effect on non-native listeners as it does on native listeners. Twenty-six adults who were native Mandarin speakers learning English as a second language were presented with English Hearing-in-Noise Test (HINT) sentences in four voice conditions with the original male speaker's fo doubled and/or VTR scaled up by a factor of 1.2: (1) low fo low VTR (LfoLVTR, the original recordings); (2) low fo high VTR (LfoHVTR); (3) high fo high VTR (HfoHVTR), and (4) high fo low VTR (HfoLVTR). The stimuli were presented in speech-shaped noise (SSN) and four-talker babble (FTB) at signal-to-noise ratios of −3, 0, +3 dB. The results showed that the non-native listeners performed more poorly with fo-VTR mismatched voices than with fo-VTR matched voices and the negative influence of mismatched voice features was mainly manifested in the HfoLVTR condition. Compared to SSN, FTB had a greater adverse impact on the non-native listeners’ recognition accuracy. Further, the performance difference between matched and mismatched conditions showed distinct patterns across SSN and FTB.