In addition to general considerations for voice talent selection in the design of an IVR (see Voice Talent), there are some situations in which there may also be accessibility considerations.

Avoid voice talents with very high-pitched voices
With coaching, you can get voice talents to adopt different rates of speaking and different amounts of dynamic variation. Voice talents, however, have limited control over the pitch of their voice.

The telephone channel is bandwidth constrained to the point that it fails to provide reliable reproduction of high-frequency speech sounds, specifically, the fricatives (e.g., /s/ vs. /f/) (Lewis, 2004). This rarely affects one's ability to understand speech over the phone because there will usually be enough additional context for listeners to accurately disambiguate the spoken message.

On the other hand, people with hearing loss have more trouble understanding higher-pitched voices (Shinn, 2009; Shinn, Basson, & Margulies, 2009). In cases where the listener has high frequency hearing loss, intelligibility issues with female and child speech are well known to audiologists, especially over the telephone (Margulies, 1980). Presbycusis, the high frequency hearing loss associated with aging, is extremely common. Self-reported hearing impairment ranges from 5.4% in 18-44 year old persons to 29.6% in those 65 and older (Schoenborn & Marano, 1988). The prevalence of measurable impairment after testing is as high as 83% in people between the age of 57 and 89 (Mościcki et al., 1985).

The available research does not provide specific guidance on how high-pitched a voice must be to consistently cause intelligibility problems for callers with hearing loss, but these data do indicate that designers should avoid selecting voice talents with unusually high-pitched voices.

Lewis, J. R. (2004). Effect of speaker and sampling rate on MOS-X ratings of concatenative TTS voices. In Proceedings of the Human Factors and Ergonomics Society (pp. 759-763). Santa Monica, CA: HFES.

Margulies, M. K. (1980). Effects of talker differences on speech intelligibility in the hearing impaired. Doctoral dissertation, City University of New York.

Mościcki, E.K., Elkins, E. F., Baum, H. M., & McNamara, P. M. (1985). Hearing loss in the elderly: An epidemiologic study of the Framingham Heart Study cohort. Ear and Hearing Journal, 6, 184-190.

Schoenborn C. A., & Marano, M. (1988). Current estimates from the national health interview survey: United States 1987. In Vital and Health Statistics, series 10, #166. Washington, D.C.: Government Printing Office.

Shinn, P. (2009). Getting persona – IVR voice gender, intelligibility & the aging. In Speech Strategy News (November, pp. 37-39).

Shinn, P., Basson, S. H., & Margulies, M. (2009). The impact of IVR voice talent selection on intelligibility. Presentation at SpeechTek 2009. Available at <www.speechtek.com/2009/program.aspx?SessionID=2386>.