Stimulus Levels for Pediatric REM

Question

What stimulus levels should be used to verify average conversational speech when performing real ear measurements in children?

Answer

Verifying hearing aid performance for average conversational speech is usually aimed at measuring the frequency shaping of the fitting for use during conversations in quiet. One such situation has been well-described in the literature: a talker uses a normal vocal effort and is directly in front of the listener, with about 1 meter spacing between talker and listener (Byrne et al, 1994, Cox & Moore, 1988, Olsen, 1998). In this situation, most people produce speech that is about 60 dB SPL, overall. If the talker is closer, or if the talker uses a raised vocal effort, higher overall levels in the range of 65 are also realistic (Olsen, 1998). An input level of 70 dB SPL, however, is closer to a "loud" vocal effort, and is likely higher than what truly reflects "average conversation". In practice, it is often useful to perform verification measures using levels that help us verify WDRC hearing aids. This suggests two things. First, it helps to verify the frequency shaping of the fitting above the WDRC compression threshold (CT), so that the shape is not changed by the hearing aid moving into and out of compression across channels. Second, it is often a good idea to verify across input levels to quantify output and audibility for a range of speech levels. A 65 dB SPL "average" speech estimate is high enough that it exceeds the WDRC CT in most cases, and not so high that it represents loud speech. However, today's generation of prescriptive targets will vary the target with the chosen verification test level, so you can feel free to choose the input level that is most helpful for each child. If you have a low WDRC CT and wish to know about "normal conversation", you can use a test level of 60 dB SPL. If you have a higher WDRC CT and wish to make sure you are above CT, you can use a test level of 70 dB SPL. Putting all of this together, we typically use 55 (soft), 65 (average), and 75 (loud) dB SPL inputs. These general concepts apply to both adults and children, at least for face to face communication. Other listening situations (such as the classroom) may require different approaches.

References

Byrne D, Dillon H, Tran K, Arlinger S, Wilbraham K, Cox R, et al. (1994). An international comparison of long-term average speech spectra. J Acoust Soc Am 96: 2108-2120.

Cox RM and Moore JN. (1988). Composite speech spectrum for hearing aid gain prescriptions. J Speech Hear Res 31: 102-107.

Olsen WA. (1998). Average speech levels and spectra in various speaking/listening conditions: A summary of the Pearson, Bennett, and Fidell (1977) report. Am J Audiol 7: 1-5.

Dr. Susan Scollie is an Assistant Professor of Audiology at the University of Western Ontario. Together with colleagues at Canada's National Centre for Audiology, she has contributed to version 5.0 of the DSL Method, and finds speech-based real ear measurement to be really useful during hearing aid fitting and troubleshooting.