Added sounds for quiet vehicles Prepared for Brigade Electronics by Dr Geoff Leventhall October 21 1. Introduction.... 2 2. Determination of source direction.... 2 3. Examples of sounds... 3 4. Addition of a musical chord... 5 5. Loudness of the demonstration sounds... 6 6. Directivity of the sounds.... 7 7. Attenuation... 9 8. And Finally... 9 Definitions Locatable (adj.): Describes a sound whose source direction is readily discernible by a listener. Locatability (noun): Locatability has two components; (1) determination of direction of a sound source and (2) estimation of its distance Directivity: The property of a source which describes how the source radiates energy in given directions. Attenuation: Describes the reduction of sound energy due to its dissipation Audibility: Describes the degree of perception of a sound by a listener 1
Added sounds for quiet vehicles Prepared for Brigade Electronics by Dr Geoff Leventhall October 21 1. Introduction. An essential element of safety in the presence of moving vehicles is the ability to locate the direction in which a nearby vehicle is travelling and also to estimate its distance. Identification of the direction is determined psychoacoustically, as described below. Estimation of the distance is a learned effect, developed from familiarity with vehicle sounds, particularly their loudness. Knowledge of direction and distance combine to give the location of a vehicle, referred to as its locatability. 2. Determination of source direction. Three overlapping frequency regions have been shown to influence the perception of the direction of a sound source. The time difference between arrival of sound at the ears. This interaural time difference (ITD) is a relatively low frequency phenomenon and operates up to about 15. 1 The sound pressure difference between sound at the two ears is the interaural intensity difference (IID) and operates from about 7 upwards but most significantly at higher frequencies. At frequencies above about 5, reflections from the outer ear into the ear canal give further information on location of a source. Determination of direction is easiest when the sound contains maximum information over all frequency ranges, which is given by white sound. However, white sound has a harsh quality: (Click the icon to hear the sound). White sound 2 and pink sound, falling at 3dB/octave is a more pleasant sounding alternative. 1 Human Localisation, Binaural cues http://www.isvr.soton.ac.uk/fdag/vap/html/localisation.html 2 ID-6 2
Pink sound 3 Consequently, where broadband sound is used, pink sound is made the basis for the following demonstration sounds, which have been developed to illustrate the variety of sounds which have good locatability. The spectra of white sound and pink sound are shown in Figs 1 and 2. The top plot is a narrow band constant bandwidth spectrum, with frequency resolution of 1. White sound has a uniform spectrum on the constant bandwidth plot, but rises at 3dB per octave on the third octave plot, due to the widening of the bands as frequency rises. Pink sound falls at 3dB/octave on the constant bandwidth plot, but is level on the third octave plot. In the spectra in Figs 1 to 9, levels are relative. The upper spectrum is a narrow band analysis with 1 frequency resolution. The lower spectrum is one-third octave. All third octave analyses have been normalised to a level of 6dBA in order to permit loudness level (phons) to be calculated and compared, as in Section 4. 3. Examples of sounds All samples are shown with constant level and frequency, although variation of these can be used to indicate vehicle speed. locatability and is not recommended by the Working Group. A single tone has poor 3.1 Two tones The simplest, but least effective, locating sound is two tones, one in the lower frequency region plus one in the higher frequency region for location. (These tones, used for Illustration, are taken from the Japanese guidelines) 3 ID 7 3
Two tones 4 6 and 2.5k The spectra are shown in Fig 3. The sound will have poor locatability. 3.2 Two frequency regions A development from two tones is to use a low and a high frequency sound region as in the following: Two frequency regions 5 covering the following third octave bands 8, 1 and 125. 25, 315, 4 In this, the spectra of Fig 4 show a fairly slow fall off away from the centre frequency of the main bands. 3.3 Two frequency regions plus tone 3.3.1 Two frequency regions plus tone 6 This is the same as in 2.2 but with a 1 tone added. The tone is about 1dB above surrounding noise on a narrow band analysis, but about 5dB above adjacent third octave bands. Spectra are in Fig 5. 3.3.2 Two frequency regions plus tone 7 As above with 25 tone instead of 1. The tone is about 2dB above the surrounding sound on the narrow band analysis, but about 5dB above adjacent third octave bands. Spectra are in Fig 6. 4 ID-5 5 ID-8 6 ID-2 7 ID-9 4
3.3.3 Two frequency regions 8 Two frequency regions, similar width to the above, but with rapid fall off at sides. Spectra are in Fig 7. 3.3.4 Two frequency regions plus tone 9 with 2 tone added. Spectra are in Fig 8. This is similar to 3.3, but 3.3.5 Two frequency regions 1 Lower frequency region is narrower than higher frequency region. Spectra see Fig 9. 4. Addition of a musical chord The demonstration sounds illustrate the range of sounds which give good locatability, whilst leaving room for personalisation by manufacturers. For example, sound 3.2 (ID-8) contains low and high frequency broadband sound, but can include either fixed or speed controlled tones within the frequency region between the two broadband sounds. The following sound 11 is ID-8 with a fixed or speed dependent chord added. The chord consists of three notes at frequencies B4 = 493.88, D5 = 587.33, F5# = 739.99. ID-8 with fixed frequency chord The frequency can be varied as in the following 8 ID-1 9 ID-3 1 ID-4 11 ID -14 5
ID-8 with variable frequency chord The narrow band spectrum for the addition of sound ID-8 and the fixed chord is at the top of Fig 1. The bottom of Fig 1 is the spectrum with frequency shifted chord. Here the rectangular box shows the extent of the shift. The lower margin is 494, the upper is 916. Thus the shift of the F5# note is from 74 to 916, a ratio of 1.24. Other tones shift ratios can be chosen. 5. Loudness of demonstration sounds ANSI S3.4 27 Procedure for the Computation of Loudness of Steady sounds provides a convenient way of calculating the loudness of sounds, in the unit phon. 4 phon is the loudness of a 1 tone at 4dB. 8 phon is the loudness of a 1 tone at 8dB etc. Consequently if a complex sound has a loudness of 8phon, it is equally loud to a 1 tone at a level of 8dB. The calculated loudness of the sample sounds is shown in Table 1 for all sounds normalised to 6dBA. The broadband sounds are close to 8phon, with the exception of sound ID-4 in Fig. 9 which has a phon level of 77.6. The two tone sound( ID-5 and Fig. 3) has a phon level of 7.5. In order to match the phon levels of the broadband sounds, the two tones have to be increased by about 1dB, giving an A-wtd level of 7dBA. Table 1 Summary of Sounds Sound Characteristic Loudness in phon at 6dBA ID -2 As ID 8 but with tone at 1k 8.1 phon ID - 3 Similar to ID 1, with tone at 2 79.3 phon ID - 4 Two humped, but lower frequency hump narrowed 77.6 phon ID - 5 Two tones equal level, 6 and 2.5k 7.5 phon 6
ID 8 Two bands of overlapping broadband 8.2 phon ID 9 As ID 8 but with tone at 25 79.5 phon ID 1 Two bands with faster fall-off than ID-8 79.2 phon It is seen from Table 1 that a good spread of broadband sound gives a good phon level, but the detail of the broadband is not too critical, as long as the spread is wide. At least an octave in both high and low frequency regions is required. 6. Directivity of the sounds. Directivity depends on two main factors: the source (loudspeaker) dimensions and the wavelength of the sound. For a fixed source dimension higher frequencies are radiated more directionally than lower ones, and concentrated forward of the vehicle. The detail of the directivity will be affected by the mounting of the loudspeaker at the front of the vehicle, but general principles still apply. Fig 11 shows the radiation patters (for source diameter/wavelength) of.25 and 1. Fig 11 Directivity patterns 7
If we consider a source diameter of.1m, representing the loudspeaker in the sounder, then the corresponding wavelengths for the two conditions in Fig 11 are.4m and.1m, or frequencies of 85 and 34. Consequently, lower frequencies spread out, whilst higher frequencies are directed forward. This means that the main sound is radiated forward of the vehicle, although lower frequencies give information, for example, to pedestrians approaching a T- junction. 6.1 Sound fields. The directional patterns of broadband and tonal sounds, when used in reversing alarms, have been determined by Laroche as in Fig 12, which compares tonal, multi-tonal and broadband sounds. It is seen that the broadband sound has the most regular radiation pattern, with highest levels concentrated down the axis of the alarm. From Fig 11, it is the higher frequencies in the sound which are concentrated axially, whilst lower frequencies spread out. Similar considerations apply to the use of broadband and tones in vehicle sounders. Fig 12. Fig 4 from Laroche Comparison of different vehicle backup-alarm types with regards to worker safety Proc ICBEN, 211 8
It should also be noted from Fig 12 that a tonal sound results in multiple reflections from nearby surfaces, causing interference effects, leading to an irregular sound field. Multiple tones even-out the sound field, but it is only the broadband sound which has clearly defined radiation characteristics, making it the optimum choice for a vehicle sounder. 7. Attenuation The attenuation of sound with distance is greatest for high frequencies, due to air and ground absorption. These two factors are not important for street traffic, but could be considered for an isolated vehicle. However, attenuation through facades adjacent to the road is very important, and also increases with frequency. This means that the higher frequencies emitted by a sounder will be at low levels within buildings, due to both the directionality of the radiation and the effect of the building façade. These result in reduced A- weighted and phon levels within the building. 8. And Finally It has been shown that a wide variety of sounds can be added to quiet vehicles. However, in order to maintain good locatability it is necessary to include broadband sounds in both low and high frequency ranges. There is a variety of choices for this. Tones, either fixed or variable, can be added to the broadband sounds. Again there is a wide variety of choice. A good spread of broadband frequencies gives a higher loudness (phon level) for a given A-weighted level of sound, compared with tones on their own. 9
-1-2 -3-4 White noise White 7 6 5 Level db 4 3 2 1 125 16 2 25 315 4 5 63 8 1. k 1.25 k Frequency 1.6 k 2. k 2.5 k 3.15 k 4. k 5. k 6.3 k 8. k 1. k dba Fig 1 White sound. Top: narrow band spectrum. Bottom: third octave spectrum. (ID 6) Series1 1
-1-2 -3-4 Pink noise Pink 7 6 5 Level db 4 3 2 1 125 16 2 25 315 4 5 63 8 1. k 1.25 k Frequency 1.6 k 2. k 2.5 k 3.15 k 4. k 5. k 6.3 k 8. k 1. k dba Fig 2 Pink sound. Upper narrow band spectrum. Bottom: third octave spectrum (ID- 7) 11
-2-3 -4-8 -9 ID5 Two tones equal amp 6 and 25 ID-5 7 6 5 Level db 4 3 2 1 125 16 2 25 315 4 5 63 8 1. k 1.25 k Frequency 1.6 k 2. k 2.5 k 3.15 k 4. k 5. k 6.3 k 8. k 1. k dba Fig 3 Two tones 6 and 25 (ID-5) 12
-1-2 -3-4 -8 ID-8 Two humps ID-8 7 6 5 Level db 4 3 2 1 Fig 4 125 16 2 25 315 4 5 63 8 1. k 1.25 k Frequency 1.6 k 2. k 2.5 k 3.15 k 4. k 5. k 6.3 k 8. k 1. k dba 8, 1, 125 bands plus 25, 315, 4 bands (ID-8) 13
-1-2 -3-4 -8 ID 2 1 tone. 8-125 and 25-4 3rd octs ID-2 7 6 5 Level db 4 3 2 1 125 16 2 25 315 4 5 63 8 1. k 1.25 k Frequency 1.6 k 2. k 2.5 k 3.15 k 4. k 5. k 6.3 k 8. k 1. k dba Fig 5 As in Fig 4 plus tone at 1, about 5dB above adjacent 1/3 oct bands. Just audible? (ID-2) 14
-1-2 -3-4 -8 ID-9 Two humps plus 25 ID-9 7 6 5 Level db 4 3 2 1 125 16 2 25 315 4 5 63 8 1. k 1.25 k Frequency 1.6 k 2. k 2.5 k 3.15 k 4. k 5. k 6.3 k 8. k 1. k dba Fig 6 Two bands plus 25 (ID 9) 15
-1-2 -3-4 -8 Two bands rapid fall off ID-1 7 6 5 4 3 2 1 125 16 2 25 315 4 5 63 8 1. k 1.25 k 1.6 k 2. k 2.5 k 3.15 k 4. k 5. k 6.3 k 8. k 1. k dba Fig 7 Two frequency regions, rapid fall off (ID-1) 16
-2-3 -4-8 -9 ID 3 2 tone. 8-125 and 25-4 3rd octs. Faster fall off than ID1/2 ID-3 7 6 5 Level db 4 3 2 1 125 16 2 25 315 4 5 63 8 1. k 1.25 k Frequency 1.6 k 2. k 2.5 k 3.15 k 4. k 5. k 6.3 k 8. k 1. k dba Fig 8 Bands at 8, 1, 125 and 25, 315, 4 with tone at 2 (ID-3) 17
-2-3 -4-8 -9 LF narrow band of noise HF wider band. LF1-122 HF 2-4 ID-4 7 6 5 Level db 4 3 2 1 125 16 2 25 315 4 5 63 8 1. k 1.25 k Frequency 1.6 k 2. k 2.5 k 3.15 k 4. k 5. k 6.3 k 8. k 1. k dba Fig 9 Two bands Lower frequency narrower than higher frequency. (ID-4) 18
-1-2 -3-4 -8 ID-8 + chord Fixed tones -1-2 -3-4 -8 ID-8 + chord frequency shift Speed dependent tones Fig 1 Spectra with fixed and speed dependent tones. Compare Fig 4 19