Experiments on tone adjustments Jesko L. VERHEY 1 ; Jan HOTS 2 1 University of Magdeburg, Germany ABSTRACT Many technical sounds contain tonal components originating from rotating parts, such as electric engines or tire/road noise. These are often considered to be more annoying than sounds without tonal components. Thus it is important to consider tonal components when assessing noise pollution. Several standards include sections dedicated to the assessment of tonal components in sound. These have in common that the magnitude of the tonal foreground is estimated relative to the noise background (e.g., tone-to-noise ratio, prominence ratio). In some standards, the higher annoyance of sounds containing tonal components is addressed by making a tone adjustment, i.e., by adding a few decibels to the measured sound level. The present study focusses on the German standard DIN 45681, where the tone adjustment increases with tone-to-noise ratio. The aim was to assess the tone adjustment in an adaptive matching experiment. First, the individual masked threshold of the tonal component was measured. Then a noise alone was adjusted to elicit the same loudness or preference as a mixture of tone and noise. The tone was added to the noise at different levels above the individual masked threshold. The results are compared to the tone adjustments calculated on the basis of the standard. Keywords: Noise immission, Preference, Loudness, Tone adjustment I-INCE Classification of Subjects Numbers: 63.2, 63.1, 63.7 1. INTRODUCTION Tonal components in environmental noise are often perceived as being more annoying / less preferred than sounds without these tonal components. Several standards assessing the noise level of a sound address this issue (1, 2, 3, 4). Common to all standards is that the level of the tone is compared to that of the background noise. Some standards request that the existence of a tonal component has to be reported if the tone level exceeds a certain threshold (1, 2). Others add a certain level to the measured level if the sound contains an audible tone (3, 4) This added level is commonly referred to as tone adjustment. The German standard DIN 45681 (4) provides tone adjustments that depend on the level of the tone above its masked threshold ( L). These tone adjustments range from 0 db (no adjustment) to 6 db and are shown in Table 1. Table 1 Tone adjustments according to DIN 45681 L in db L 0, 0 L 2, 1 L 4, 2 L 6, 3 L 9, 4 L 12 5 L 6 Tone adjustment in db 1 jesko.verhey@med.ovgu.de 2 jan.hots@med.ovgu.de 4538
There is a compressive relation between tone adjustment and L. The present study investigated the relation between the tone level L and the preference in normal-hearing listeners. To this end, the level at equal preference between a noise-only stimulus and the noise where a tone was added were measured for three target-tone frequencies. The levels of the tone were chosen relative to the individual masked thresholds of the tones that were measured prior to the preference experiment. The sensation associated with the perception of audible tones in noise is referred to as tonality, tonalness or magnitude of tonal content (5). Recent studies indicate that this sensation is closely related to the sensation of the partial loudness of the tonal component (6, 7). The change in loudness due to the presence of the tonal component may also account for the change in preference. To test this hypothesis, for one tone frequency, the level of a noise-only stimulus was measured that was perceived as equally loud as a combination of the tone and the noise. 2. METHODS 2.1 Listeners Ten normal-hearing listeners aged between 20 and 35 years participated in the experiments. Their thresholds in quiet were 15 db for all audiometric frequencies between 125 Hz and 8000 Hz. Seven listeners had prior experience with psychoacoustic experiments. All listeners were paid voluntee rs. 2.2 Apparatus The listeners were seated in a double walled sound insulated booth. The signals were generated digitally in Matlab at a sampling rate of 44100 Hz, transformed to analog signals with an external sound card (RME Fireface 400) and presented to the listeners via headphones (Sennheiser HDA 200). The headphones were free-field equalized according to DIN EN 389-8 (8). 2.3 Stimuli All stimuli contained a uniform exciting noise (UEN). In some stimuli a pure tone was added. The frequency and level of the tone varied across experiments. UEN and tones were switched on and off simultaneous. Stimulus onsets and offsets were tapered with 50-ms von-hann windows to prevent audible acoustic transients. 2.4 Procedure The study consisted of three experiments. Their procedures are described in the following three subsections. 2.4.1 Masked threshold experiment Masked thresholds were measured for target frequencies of 350 Hz, 700 Hz and 1400 Hz with an adaptive 3-interval 3-alternative-forced choice (AFC) procedure. The intervals of a trial were separated by periods of silence lasting 500 ms. In each interval of a trial, a 60-dB UEN was presented and in one randomly chosen interval the target tone was added. The task of the listener was to indicate this target interval. The level of the target in the first trail was 63 db SPL. For the next trials, the level of the target was varied adaptively using a one-up two-down rule. The step size was 6 db at the beginning of a run, reduced to 3 db after the first upper reversal and to 1 db after the second upper reversal. With this minimum step size the run continued for another six reversals. The mean of the levels at these reversals was taken as an estimate of masked threshold. The run was repeated three times and the mean of the three estimates were taken as the final estimate of the masked threshold. 2.4.2 Preference experiment The tone adjustments of the German standard DIN 45681 (4) reflect the lower preference of a sound containing a tone compared to a similar sound without tonal components. This experiment investigated which level of a noise only stimulus was equally preferred to a noise where a tone was added. A 2-interval 2-AFC procedure was used to measure this level at equal preference for three different target levels. The target levels ( L) were 5 db, 10 db and 15 db above masked threshold of the listener. 4539
Target frequencies were 350 Hz, 700 Hz and 1400 Hz. The stimulus intervals were 2000 ms long and were separated by 500 ms of silence. One interval contained the target and the UEN (target interval), the other interval contained the UEN alone (reference interval). The task of the listeners was to decide which of the intervals they preferred. The level of the UEN in the reference interval was varied adaptively using a one-up one down rule. Three starting level of the UEN in the reference interval were used: 50 db, 60 db and 70 db SPL. The initial step size was 8 db and was halved after each upper reversal until the minimum step size of 2 db was reached. The track continued for another 4 reversals. The mean of these four reversals was taken as estimate of the level at equal preference of the noise-only stimulus. As a final estimate for each frequency-level combination of the target the levels at equal preference for the three starting levels were averaged. For each target frequency, the tracks for the three starting levels and three target levels were interleaved to reduce bias effects (see, e.g., 9). The runs for the different target frequencies were randomized for each listener. 2.4.3 Loudness experiment For one target frequency (700 Hz), the loudness of the noise -only stimulus was adjusted to the loudness of the mixture of tone and noise at the three target levels 5 db, 10 db and 15 db above masked threshold of the listener. In general, the experimental paradigm was the same as used in the preference experiment. The only difference was that the listener was now asked to decide which of the two intervals of a trial the louder one was. 3. Results and Discussion Figure 1 shows level differences between the equally preferred UEN and UEN with the added tone as a function of the level L of the tone above its masked threshold. The symbols indicate the frequency of the target tone. All level differences are positive, i.e., a combination of tone and noise (UEN+tone) requires a lower level than the noise-only stimuli (UEN) in order to be equally preferred, as expected. Figure 1 Level difference between uniform exciting noise (UEN) alone and noise with an added pure tone at equal preference as a function of the level of the tone above masked threshold ( L). The mean of the ten participants is shown. Error bars indicate the interindividual standard deviations of the mean. The tone frequencies were 350 Hz (circles), 700 Hz (squares), and 1400 Hz (diamonds). Horizontal lines mark the tone adjustments according to DIN 45681 for the corresponding L. For the 1400-Hz tone, the level difference increases from about 3 db for a L of 5 db to 5 db for a 4540
L of 15 db. The level differences are slightly smaller for the lower frequencies. The thick horizontal lines indicate the tone adjustment according to DIN 45681 (see Table 1). In general, these tone adjustments are in the range of the interindividual standard deviations of the level differences at equal preference of the present study. Thus, our data supports the compressive relation of tone level and tone adjustment of the German standard. Figure 2 shows the results of the loudness experiment (open symbols). In addition, the corresponding results of the preference experiments are redrawn from Figure 1 (filled symbo ls). The level differences at equal loudness tend to be smaller than those for the preference experiment. A similar result was obtained in a study with multi-tone complexes (10), indicating that changes in overall loudness due to the addition of the tone may contribute to the higher annoyance of the sound but are presumably not accounting for the whole effect. Figure 2 Mean level differences and standard deviations between UEN and UEN with an added 700-Hz pure tone at equal preference (gray squares) and at equal loudness (open squares) of the ten participants as a function of L. The corresponding tone adjustments according to DIN 45681 are shown with horizontal lines. The compressive relation between tone adjustment and the level above threshold may reflect the nonlinear change in tonalness with level (6). Since partial loudness of the masked tone changes in a similar way, future standards may consider using the perceptually more relevant partial loudness of the tonal content instead of the level of the tonal component above thresholds to derive the tone adjustment. 4. CONCLUSIONS Sounds with a tonal content are more annoying than those without these components. Tone adjustments were proposed that account for this effect. The present study showed that measured level differences between a noise and a noise with an added tone are very similar to the tone adjustments of the German standard DIN 45681. The increase in overall loudness when the tone is added to the noise is presumably only accounting for part of the increase in annoyance. It remains to be seen if the perceptual measure partial loudness of the tonal content may be better related to the increase in annoyance than the more physical measure level above masked threshold. 4541
ACKNOWLEDGEMENTS This work was supported by the Deutsche Forschungsgemeinschaft (DFG, SFB trr 31). REFERENCES 1. ANSI S1.13 Measurement of Sound Pressure Levels in Air. 2005 2. IEC 61400-11 Wind turbine generator systems - Part 11: Acoustic noise measurement techniques Reference. 2006 3. ISO 1996-2 Acoustics - Description, measurement and assessment of environmental noise - Part 2: Determination of environmental noise levels. 2007 4. DIN 45681 Acoustics - Determination of tonal components of noise and determination of a tone adjustment for the assessment of noise immissions. 2005 5. Hansen, H., Verhey, J.L., Weber, R. The magnitude of tonal content. A review. Acta Acustica united with Acustica 2011; 97:355-363 6. Verhey, J.L., Heise, S.J. Suprathreshold perception of tonal components in noise under conditions of masking release. Acta Acustica united with Acustica 2012; 98:451-460 7. Hansen, H., Weber, R. Partial loudness as a measure of the magnitude of tonal content. Acoustical Science and Technology 2010; 32:111-114 8. DIN EN ISO 389-8 Acoustics- Reference zero for the calibration of audiometric equipment - Part 8: Reference equivalent threshold sound pressure levels for pure tones and circumaural earphones (ISO 389-8:2004); German version EN ISO 389-8:2004 9. Verhey, J.L., Kollmeier, B. Spectral loudness summation as a function of duration. Journal of the Acoustical Society of America 2002; 111(3): 1349-1358 10. Töpken, S., Verhey, J.L., Weber, R. Preference and loudness of multi-tone sounds. Proc. AIA-DAGA 2013; 18-21 March 2013; Merano, Italy 2013. p. 1269-1272 4542