The change in overall level, spectral shape, and loudness perception of speech produced with different vocal effort

New prescriptive procedures are required for fitting nonlinear hearing aids that prescribe both frequency and level dependent gain. Many established fitting procedures for linear hearing aids use a representation of the speech spectrum in the rationale to prescribe gain and the frequency response shape, and speech is also likely to play an important role in coming rationales for fitting nonlinear hearing aids. Therefore, we need to know more about how the behaviour and the perception of speech change with overall level. This paper presents overall levels and spectra of ten female and ten male voices produced with Soft, Casual, Normal, Raised, and Loud vocal efforts. Individual recordings were conducted in an anechoic chamber. The overall levels span about 20 dB from Soft to Loud, and the relative difference between spectra showed that less energy was produced in the low frequencies than in the mid frequencies as the vocal effort increased. A set of combined speech spectra for Soft, Casual, Raised, and Loud vocal efforts, referenced to the International long-term average speech spectrum (ILTASS) is presented. An alternative Soft speech spectrum that takes both soft voices and distant talkers into consideration is also included. Finally, loudness growth functions for wideband speech shaped to match the Loud, Normal and Soft/distant spectra were measured. The difference in spectral shape did not influence loudness perception in normal ears. Implications of the findings for hearing aid fitting are discussed.