Comparison of three calibration procedures for free-field reference speech audiometry

Abstract

Objective

In standards IEC 60645-1 and ANSI S3.6, the free-field equivalent earphone output level method is assumed as the reference for speech audiometry. Three calibration procedures for this method were compared in this study.

Design

Speech audiometry was conducted with Dutch consonant-vowel-consonant words for the following conditions: 1. TDH39 earphones, 2. loudspeaker, and 3. free-field simulated with TDH39 earphones. The first calibration procedure was based on the empirically determined difference between the speech recognition threshold (SRT) with earphones and a loudspeaker. The second procedure was based on the theoretical free-field correction, derived from the known speech spectrum and the free-field to coupler difference. The third calibration procedure corresponded to the results of the free-field simulated speech material under earphones.

Study sample

The sample included 20 normal hearing subjects.

Results

The differences between the observed SRT in the free-field and earphone conditions and the free-field and simulated free-field conditions were 7.1 dB and 0.6 dB, respectively.

Conclusion

The three calibration procedures for the free-field equivalent output method yielded approximately the same results, and therefore all appear to be useful for TDH39 earphones.

Keywords:

• Calibration
• speech audiometry
• free-field equivalent output method
• speech intelligibility index
• coupler to free-field difference

Acknowledgments

The authors wish to acknowledge Professor Dr. Ir. W.A. Dreschler for reviewing the manuscript of the first edition and offering valuable comments and suggestions. Hans Versluis is acknowledged for his textual comments. The authors thank the normal hearing volunteers who took part in this study.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 The NVA (Nederlandse Vereniging voor Audiologie) is the Dutch Association of Audiology.

2 These word lists were only validated for TDH39 earphones with normal hearing subjects by determining the relationship between the subject’s response (i.e. phoneme and word recognition scores of monosyllables in percentages) and the magnitude of a physical stimulus (i.e. speech level in sound pressure level). Only the phoneme recognition scores are used in Dutch clinical practice. The following reference characteristics were determined by Bosman and Smoorenburg (1992): SRT of 20.7 dB SPL and slope of 5.1%/dB for phoneme scores. For word scores a SRT of 20.7 + 4.7 = 25.4 dB SPL and an equal slope to the phoneme scores of 5.1%/dB can be derived (Bosman, Wouters, and Damman 1995; Figure 7A).

3 Two dB of the 10 dB discrepancy is due to the monaural-binaural difference (French and Steinberg 1947; Killion 1978; Brinkmann and Richter 1997).

4 RETSPL (rel. 1kHz) monaural free-field speech = 12.5dB + 2.4 (1 kHz RETSPL) + 2.0 dB (binaural to monaural correction; Killion 1978) = 16.9 dB SPL; RETSPL (rel. 1kHz) monaural TDH39 Speech (coupled in a IEC 60318-1 artificial ear) = 12.5 dB + 7.5 (1 kHz RETSPL) = 20.0 dB SPL; difference RETSPL Speech = 20.0 dB SPL – 16.9 dB SPL= 3.1 dB.

5 Loudness comparison tests are performed with a group of otologically normal subjects. Their task is to compare the loudness of two narrow-band noise signals: one presented by a loudspeaker (calibrated in the free-field) and the other presented by an earphone under test (calibrated on an artificial ear, e.g. B&K 4153; IEC 60318-1 coupler). The subjects adjust the signal level of the earphone under test until it is equal to the loudness perception of the signal presented by the loudspeaker.