Speech spectral changes at extended high frequencies associated with increased vocal effort

Until recently, it has been assumed that extended high frequencies (EHFs; > 8 kHz) in speech do no contribute to speech perception, being beyond the traditional “speech bandwidth.” However, EHFs in speech provide cues beneficial for speech-in-noise listening. This is due, in part, to the fact that EHFs in a target talker’s speech tend to be higher in level than EHFs in noise or competing speech for real-world settings. To be heard in noisy situations, talkers increase their vocal effort, resulting in changes to the speech spectrum. Spectral changes associated with increased vocal effort have been examined previously, but these analyses have not included EHF energy. In the present study, we examined spectral changes associated with increased vocal effort across the entire audible speech spectrum for trained vocalists (N = 15; 8 female). We observed fundamental frequency increases and shifts in spectral energy towards 2-4 kHz as others have reported, although these effects were not universal. Absolute levels of EHF energy universally increased. EHF relative levels stayed the same for approximately half of the vocalists and were reduced for the others. Whether these spectral changes affect the utility of EHFs for speech-in-noise recognition remains to be seen.