INTER-NOISE 2007 28-31 AUGUST 2007 ISTANBUL, TURKEY The acoustic description of patterns in soundscapes Klaus Genuit a André Fiebig b HEAD acoustics GmbH Ebertstrasse 30a 52134 Herzogenrath GERMANY ABSTRACT Urban places can be characterized by acoustical fingerprints, being composed of specific sound sources dynamically emerging at certain times. This results in specific noise features and patterns of the soundscape, which allow to identify the urban place simply by hearing. Experiments have shown that listeners are able to recognize places or at least to assign heard environmental sounds to cultural areas. This means that the continually varying acoustical scenery offers cues, such as specific noise components, noise patterns or temporal information, which can reliably be interpreted. Therefore, essential sound features of given urban sound scenarios were exemplarily determined on the basis of Schafer s main themes of soundscapes key notes, signals and sound marks. The study deals with the question whether it is possible to identify these attributes by means of acoustical analyses. In particular, benefits and limits of common acoustical parameters, such as psychoacoustic parameters, concerning the classification of soundscapes will be illustrated for a few given soundscapes. The findings will be discussed with regard to the application of the soundscape approach to noise policy. 1 INTRODUCTION The increasing number of recent publications about soundscape models and approaches is evidence of the growing insight into the complexity of environmental noise perception and evaluation. Several studies have proved that noise annoyance based on sound pressure level values alone cannot be reliably determined. [1, 2, 3] Murray Schafer already recognized this more than thirty years ago. He demanded for tools and methods to discover the significant features of the soundscape, those sounds which are important either because of their individuality, their numerousness or their domination. [4] However, community noise and soundscape researcher are quite far from attaining this goal. Schafer s defined main themes of soundscapes key notes, signals, sound marks, and archetypal sounds are, up to now, more theoretical and qualitative-descriptive categories than systematically specified hearing events. A hearing event is developed in the listener s ears, it is a cognitively processed physical event. The physical event is processed as a meaningful event based on source identification effects. [5] Nevertheless, similarities between sound events and resulting evaluations need to be investigated. In particular with respect to noise policy, such findings could be valuable regarding the determination of soundscape quality, especially in cases, where extensive a Email address: klaus.genuit@head-acoustics.de b Email address: andre.fiebig@head-acoustics.de
surveys and questioning of residents are not realizable. In the following, the possibility of detecting physical clues referring to soundscape themes will be exemplarily investigated. The proposed soundscapes themes by Schafer are understood as tentative categories. 2 CLASSIFICATION OF SOUNDSCAPES According to our experience, cities around the world show differences in how they sound. Due to cultural, social, geographical, meteorological, and other conditions a city has its own specific sound, its characteristic music and melody. However, in most cases it is difficult to systematically explain the particularities of a soundscape. In the following, comparisons between cities from different continents are exemplarily carried out to underline the uniqueness of soundscapes. It has to be remarked that the recordings cannot be regarded as representative for the whole city because of the variety of soundscapes within a city, and because the recordings do not capture all features of the soundscapes due to the limited measurement time. However, the following considerations illustrate the soundscape differences, which can be (partially) characterized in terms of acoustical descriptions. For example, a well known soundscape with its specific sound is the American city New York City. The town (only Manhattan is considered here) is shaped by cab horns, low frequent noise caused by air conditioning systems, squeal brake noise, police and fire engine sirens, etc. It forms a unique soundscape which can be identified simply by hearing. Figure 1: New York City, 5 th Avenue and Broadway (USA) In contrast to it, the soundscapes of European cities are often dominated by vehicles tire and power train noise. Furthermore, in several countries motorcycles and motor scooters are widely spread; they considerably influence environmental noise. For example, 35% of all motorcycles in Europe can be found in Italy, where most of the motorbikes possess less than 50cc. [6] This has a great impact on the soundscapes in Italian cities. Furthermore, in many cities small power motors, mostly four-cylinders, determine the environmental noise. In contrast, in other cities a more complex traffic composition with lots of cars with six and eight cylinders (luxury cars) can be found. In the following, two European examples are studied, the city of Paris, Avenue du Général de Leclerc and the city of Pisa, Arno Waterside. Figure 2: Left: Paris, Avenue du Général de Leclerc (France), Right: Pisa, Arno Waterside (Italy)
In Japan, low frequency contributions are less important in contrast to the USA, where theyare often found due to numerous HVAC systems. Furthermore, traffic noise in Japanese cities is often not dominated by motorcycles, but rather determined by sedans and limousines. Moreover, in Japan cities, as Hiramatsu stated [7], signal sounds have become an important factor of the soundscapes, both in individual and public spaces. As an example, a recording from the city of Osaka is analyzed and compared with the other cities. Figure 3: Osaka, Japan In the south-eastern part of Asia many motorcycles and especially motor scooters can be found, which significantly shape the environmental noise. Those means of transportation are very dominant and greatly influence the soundscape. Here, a recording made in Taipei (Taiwan) is considered and analyzed. Figure 4: Taipei, Taiwan These short descriptions reflect the particularities, cultural differences and peoples preferences, which finally lead to specific environmental noise settings. However, these descriptions are quite removed from being representative and are not exhaustive. Therefore, the following discussion of characteristic soundscape features should be understood as a case study, which shows the potential of the soundscape approach. The consideration of the L Aeq, which is usually used in noise maps, is excluded from the analysis. 3 RHYTHM, TEMPO, SPECTRAL CONTENT AND PATTERNS IN SOUNDSCAPES The terms rhythm and tempo are frequently used by Schafer and refer to the [division] of the whole into parts. Furthermore, the expression pattern is a meaningful part of Schafer s soundscape theory because man is an anti-entropic creature; he is a random-to-orderly arranger and tries to perceive patterns in all things. [4] Thus, the above mentioned urban city recordings are studied and analyzed with respect to these concepts.
As Figure 5 illustrates, the low frequency contribution in the examined urban settings varies considerably. In particular, the low frequency content plays a decisive role in New York City as well as in Pisa, whereas the low frequency contributions in the other cities are not of great importance. Figure 5: Variable Frequency Resolution vs. Time. From left to right: NYC, Paris, Pisa, Osaka, Taipei The closer examination of the environmental noise recorded in New York City shows that distinct harmonics are present in the soundscape. In contrast to the city of Pisa, these tonal contributions are permanently present, whereas in Pisa the low frequency contribution caused by the traffic is more unsteady. Figure 6 displays the frequencies of the detected harmonics 30Hz, 45Hz, 60 Hz, 120Hz and 180Hz which are harmonics of the power frequency of 60Hz by which the numerous HVAC systems are operated. Figure 6: Variable Frequency Resolution vs. Time (linear frequency scale). NYC Fastl and other authors have found that the 5% percentile loudness (N 5 ), the value which is reached or exceeded in 5% of the measurement time, correlates highly with the perceived
total loudness valid in cases of unsteady sounds. [8, 9] This finding is implemented in the draft of the new DIN 45631, which defines: since the mean value of time varying loudness compared with the subjectively evaluated loudness provides a value, which is too low, the 5% percentile loudness (N 5 ) has to be used with respect to the perceived overall loudness. Further loudness percentiles can be additionally used. [10] Figure 7 shows the different loudness percentile values measured at the different locations, where longer measurement time periods are considered than depicted in figure 5. The background loudness N 95 is almost equal at all places except for Taipei and New York City, where the background loudness is very high. The loudness N 5 varies between the cities in the range of 107 sone to 50 sone. It can be seen that NYC has a high loudness value, similar to Taipei, which shows the highest loudness. The locations in Osaka, Paris and Pisa show much lesser loudness values. Time-Varying Loudness [Sone] 100 80 60 40 20 Loudness Percentiles 0 New York City 5th Avenue Paris Avenue du General Leclerc Pisa Arno Waterside N5 N50 N95 Osaka Road Figure 7: Comparison of Loudness Percentile (N 5, N 50, N 95 ) of five places Taipei Road Moreover, it was suggested that relative changes (e.g. loudness fluctuations) can highly influence the sound evaluation. The higher these fluctuations are the more annoying a sound is possibly perceived. [11] Human hearing adapts to steady signals, but remains very sensitive to fluctuations as well as to prominent, salient noise events. Therefore, peak values and relative changes can be important. To grasp these fluctuations and variations the calculation of percentiles (here loudness) as well as the calculation of percentile value differences can be a simple solution (e.g. X 5 -X 50 ). Figure 8 shows difference values between N 5 and N 50 for the examined locations. It is clearly observable that e.g. Pisa has strong loudness variations, although the loudness is not very high. This is also indicated by the ratio N 5 and N 50 ; a value of 50% means that N 5 to N 50 decreases of 50% (N 50 is half the loudness of N 5 ). New York City possesses the lowest difference value, which implies that the loudness curve over time is not strongly varying. However, in this case it has to be remarked that NYC has loud background noise and lots of noise sources are permanently present in the soundscape, so that outstanding noise events are only occurring in case of really loud events. Pisa shows a strong variation of the loudness over the measurement time (fig. 9). Several prominent noise events as well as periods of quietness are identifiable. The presented variations are an important feature of a soundscape and should be interpreted with respect to the specific character of the soundscape.
45 40 35 [Sone] and [%] 30 25 20 15 10 5 0 New York City 5th Avenue Paris Avenue du General Leclerc Pisa Arno Waterside Osaka Road Taipei Road N5-N50 [sone] Relative change N5 to N50 [%] Figure 8: Comparison of Loudness Percentile Difference (N 5 -N 50 ) and Ratio of N 5 to N 50 of five places Studies have shown that physiological responses (e.g. finger pulse amplitude, skin conductance level, electro-myogram) do not linearly change with decreasing traffic density and reduced sound pressure level respectively. [12] For traffic noise scenarios, where passby noise events are separately perceived in the midst of silence, the physiological reactions are possibly caused by startling reactions because of the non-stationary character of the noise. Such phenomena can also be expected for the examined locations. As figure 9 shows, quieter parts are interrupted by single noise events, which remarkably stand out of the background noise ( keynote sounds ). This phenomenon is also observable to a certain degree with respect to the other test sites. Figure 9: Pisa Arno Waterside: Loudness vs. time 3.1 Consideration of Typical Noise Sources within the Soundscapes As mentioned in chapter 2, typical and characteristic noise sources within the urban locations are shaping the respective soundscapes; they leave their characteristic mark on the soundscapes. Thus, it is necessary to consider and analyze the noise events caused by specific noise sources in detail. For example, motorcycles and motor scooters can be found in many Asian cities as well as in European cities, often in south Europe. The motor scooter noise
strongly influences the character of the soundscape. It is often equipped with a two-stroke engine, which possesses a sound, that frequently leads to an unpleasant sound impression and a feeling of obtrusiveness. Figure 10 illustrates the specific sound character of a motor scooter, which displays engine orders almost over the whole frequency range. In particular, in the medium and higher frequency range the engine order harmonics are strongly audible. Figure 10: Variable Frequency Resolution vs. Time. Paris Another interesting specific noise event within the considered urban sites is the police siren in New York City. This noise event can be understood as a signal after Schafer s theory of main soundscapes themes, which acts as an acoustic warning. Moreover, it is almost perceived as a soundmark, which possesses a unique quality and is interpreted, especially in combination with ceaseless honks made by taxicabs, as characteristic for New York City. This was confirmed through listening tests with naive persons. That is particularly surprising, since these sounds are not only used by the NYC police. However, the constancy of these noises, sirens and honks in the environmental noise are often interpreted as typical for the New York City soundscape. Figure 11 shows the acoustic signature of the police siren recorded in New York City. Figure 11: Variable Frequency Resolution vs. Time. New York City Another difference within the soundscapes is caused by the different traffic composition, as already mentioned above. In Europe, only ten to fifteen per cent of passenger cars are equipped with manual transmission (MT), whereas in Japan and USA the majority of
passenger cars are equipped with automatic transmission (AT) (fig. 12). This difference has already an effect on the environmental noise. Figure 13 shows a comparison between two pass-by noises one vehicle equipped with AT, one equipped with MT. It can be clearly seen that due to the respective transmissions, regardless of the differences in the engines (6 to 8 cylinders), the behavior of the engine orders is different and results in different sound impressions. The converter in the AT leads to a distinct run of the engine orders compared with the manual transmission. 100 90 80 70 60 [%] 50 40 30 20 10 0 North-America Europe Asia Japan AT (2005) AT (2010) MT (2005) MT (2010) Figure 12: Distribution of transmission types (AT=Automatic Transmission, MT=Manual Transmission) for 2005 and 2010 [13] Another relevant aspect concerns the diesel and petrol engines, which occur in different quantities in the examined sites. For example, in Germany 81.5% are petrol and 18.4% are diesel engines - comparable with the USA-, whereas in France, Belgium or Austria over 40% diesel engines can be found. [6] Such different distributions concerning the number (ratio) of diesel and petrol engines - or alternative drives - lead to a characteristic traffic noise. The specific acoustic signatures of these different engines are illustrated in figure 14 with the help of the Relative Approach analysis. Figure 13: Comparison of noise from an accelerated car equipped with manual transmission with an accelerated car equipped with automatic transmission
3.2 Consideration of Short-Term Frequency and Time Patterns in Noise Besides the number of noise events within a long-term period, short-term patterns in noise can also arouse attention and can influence the evaluation of environmental noise. The human ear is very sensitive to noise patterns and is able to identify these patterns. It is assumed that human hearing creates for its automatic recognition process a running reference sound (an anchor signal ) against which it classifies tonal or temporal pattern information moment-bymoment. Temporal structures and patterns as well as patterns in the frequency domain are important with respect to the resulting sound impression. Thus, a procedure has been developed which allows for the detection of aurally-relevant short-term patterns in noise. It is based on relative level variations within a given critical band over time, or within a time window in the frequency range related to formation of a sliding mean value. [14] An example is depicted below, where a petrol engine and a diesel engine is compared. (fig. 14) Within the European research project QCity, which is dealing with environmental noise issues, it was found that this relative parameter possesses significance regarding the description of single pass-by noise annoyance. [15] Figure 14: Relative Approach Analysis 3D: Comparison of a diesel engine (left) and a petrol engine noise (right) 4 CONCLUSIONS The soundscape approach can help to bridge gaps in the field of community noise research. It offers the possibility to integrate diverse aspects, which have an impact on the perception of environmental noise, into a broad, comprehensive concept. Altogether, social, cultural, meteorological, geographical, architectural, and other features of an urban place moderate the perception and evaluation of environmental noise. Moreover, the meaning of certain noise sources within a soundscape influences the evaluation of their noises, too. In the present paper, the acoustical properties of soundscapes are studied which alone do not allow a complete understanding of the soundscape perception and to retrace the resulting reactions of the soundscape residents. Nevertheless, it is inevitable to thoroughly study the acoustical properties and characteristics of a soundscapes as an important part of the soundscape phenomenon, because it can provide the basis or the starting point for the classification of soundscapes. Several acoustical properties were identified, which are typical for the examined urban locations. The environmental noise differs in character, time structure, variations, etc. These acoustic particularities and noise features within the different soundscapes have to be identified and
investigated. Therefore, macroscopic and microscopic analyses are needed, on the one hand, to capture the global sound impression created by the soundscape and, on the other hand, to recognize and interpret single noise events adequately, which cause strong reactions and feelings, whether positive or negative. 5 REFERENCES [1] B. Berglund, Theory and method in perceptual evaluation of complex sound, In: Fastl, H., Kuwano, S., Schick, A. (ed.). Recent Trends in Hearing Research. Oldenburg: BIS Verlag, 1996 [2] K. Genuit, Beyond the A-weighted level, Proceedings of Inter-Noise 06, Honolulu, HI, USA. [3] P. Schomer, Alternative methods to A-weighting for environmental noise assessment, Proceedings of INTER-NOISE 02, Dearborn, MI, USA. [4] R.M. Schafer, The soundscape: our sonic environment and the tuning of the world, Rochester, Destiny Books, 1977 [5] D. Dubois, C. Guastavino, M. Raimbault: A cognitive approach to urban soundscapes: using verbal data to access everyday life auditory categories, Acta Acustica united with Acustica, Vol. 92 (2006), No. 6, 865-874, S. Hirzel Verlag [6] H. Strehlow, Personenverkehr in der Europäischen Union, Verkehr 09/2006, Eurostat, ISSN 1562-1332, KS-NZ-06-009-DE-N, Europäische Gemeinschaften, 2006 [7] K. Hiramatsu, A review of soundscape studies in Japan, Acta Acustica united with Acustica, Vol. 92 (2006), No. 6, 857-864, S. Hirzel Verlag [8] I. Stemplinger, G. Gottschling, Auswirkungen der Bündelung von Verkehrswegen auf die Beurteilung der Globalen Lautheit, 23. Deutsche Jahrestagung für Akustik, Fortschritte der Akustik, DAGA 1997, Kiel, Germany [9] H. Fastl, E. Zwicker, Psychoacoustics. Facts and Models, 3. Edition, Springer Verlag, Berlin, Heidelberg, 2006 [10] DIN 45631/A1 (Version 13), Berechnung des Lautstärkepegels und der Lautheit aus dem Geräuschspektrum Verfahren nach E. Zwicker Änderung 1: Berechnung der Lautheit zeitvarianter Geräusche, May 2007 [11] K. Genuit, Acoustical parameter for the sound quality of IT-products. Proceedings of Euronoise 2006, Tampere, Finland [12] G. Notbohm, S. Schwarz, Evaluation of sound quality of urban traffic noise by psychophysiological methods, 6 th International Symposium Transport Noise and Vibration, June 2002, St. Petersburg, Russia, 2002 [13] ATZ 07-08/2006, Jahrgang 108, Vieweg Verlag, GWV Fachverlage GmbH, Wiesbaden [14] K. Genuit, Objective evaluation of acoustic-quality based on a Relative Approach, Inter-Noise 1996, Conference Proceedings, Liverpool, England, 1996 [15] EU Project Quiet City Transport (QCITY), TIP4-CT-2005-516420 (www.qcity.org)