Acoustic vehicle alert system

Abstract

The method of designing an AVAS audio signal comprises synthesising an audio signal for the AVAS system using one or more synthesis methods; determining a set of psychoacoustic metrics based on the sy

Description

TECHNICAL FIELD The present invention relates to a method of designing an audio signal for an acoustic vehicle alert system. Aspects of the invention relate to a method, a computer program product, a compute readable medium, an audio signal, an acoustic vehicle alert system, and a vehicle. BACKGROUND TO THE INVENTION Electric vehicles (EVs) and hybrid electric vehicles (HEVs) produce significantly less noise during driving than traditional internal combustion engine (ICE) vehicles. The reduced noise production, and subsequent reduction in noise pollution, is generally considered a major advantage of EVs and HEVs when compared to traditional ICE vehicles. However, the difference in noise production is particularly apparent at low driving speeds, and when navigating surroundings with other road users, such as pedestrians and cyclists. EVs I HEVs are therefore known to include acoustic vehicle alert systems for outputting sounds to highlight the presence of the vehicle to other road users in such driving conditions. In some territories, AVASs are required by EV I HEV legislation. The sounds output from systems are generally considered sub-optimal though, for example in terms of their effectiveness at alerting other road users to the presence of the vehicle, and / or in terms of the discomfort or annoyance experienced by other road users. It is against this background that the present invention has been devised. SUMMARY OF THE INVENTION According to an aspect of the invention, there is provided a method of designing an audio signal for an acoustic vehicle alerting system. The method comprises: synthesising an audio signal for the acoustic vehicle alerting system using one or more synthesis methods; determining a set of psychoacoustic metrics based on the synthesised audio signal; calculating first and second sound quality metrics based on the determined set of psychoacoustic metrics; comparing the first and second sound quality metrics to respective thresholds; and validating the proposed audio signal for the acoustic vehicle alerting system in dependence on the comparison to the thresholds. The set of psychoacoustic metrics comprises: a loudness; a sharpness; a fluctuation strength; and an articulation index. The first sound quality metric is an annoyance level of the proposed audio signal calculated based on a rational function of at least the determined sharpness and loudness divided by the determined articulation index. The second sound quality metric is a preference level calculated based on a polynomial function of at least the determined loudness and fluctuation strength. In this manner, the method is able to determine whether an audio signal will be suitable for use in an acoustic vehicle alerting system (AVAS) through an analysis of the physical properties of the audio signal. Advantageously, the method allows for rapid AVAS signal design and refinement without the time consuming and expensive subjective aspects of jury testing. In an example, the rational function may take the form of: . (AxSharpness)x(BxLoudness2) Annoyance = ---------—--------- {CxArticulation Index) where A, B, and C are numerical constants. Optionally, A, B, and C are each equal to 1. The annoyance level may, for example, be determined based on: a maximum value of the determined sharpness of the proposed audio signal; and / or a k-th percentile of the determined loudness of the proposed audio signal, k may be a positive integer greater than or equal to 75. Optionally, k is a positive integer greater than or equal to 90. The rational function may, for example, be determined based on a regression analysis of: listener annoyance values indicated by jury test data obtained for a set of test signals and the sharpness, loudness, and articulation index determined for each test signal. Each test signal may be a respective audio signal for an acoustic vehicle alerting system. In an example, the linear polynomial function may take the form of: Preference = (D x Loudness) + (E x Fluctuation strength) — F, where D, E, and F are numerical constants. Optionally, D is equal to 0.19, E is equal to 0.409 and F is equal to 1.885. The polynomial function may, for example, be determined based on a regression analysis of: listener preference levels indicated by jury test data obtained for a set of test signals and the loudness and fluctuation strength determined for each test signal. Each test signal may be a respective audio signal for an acoustic vehicle alerting system. Optionally, synthesising the audio signal comprises setting a prescribed range for each of the following sound characteristics: a frequency spectrum; a sound pressure level; and an amplitude modulation. In an example, synthesising the proposed audio signal comprises: setting the frequency spectrum of the audio signal to include at least 65% of the energy and / or sound pressure level (SPL) at frequencies between 500 Hz and 1000 Hz; and setting the amplitude modulation of the audio signal less than or equal to 10 dBA. Optionally, at least 90% of the energy and / or sound pressure level (SPL) of the frequency spectrum may be set at frequencies between 500 Hz and 1000 Hz. In an example, the frequency spectrum of the audio signal may be set to include a zero or negligible energy and / or sound pressure level (SPL) at frequencies greater than 1200 Hz. The specific ranges provided above have been found to be particularly advantageous, at least in that the audio signals synthesised are suited to the industry standards and / or requirements in countries, such as the United Kingdom. Moreover, these ranges have been found to provide audio signals which are found to be less annoying and preferable by members of a jury. The audio signal may, for example, be designed for a forward driving condition. In which case, the amplitude modulation of the audio signal may be set between 3 dBA and 6 dBA. In an example, the audio signal may be designed for a reverse driving condition. In which case, synthesising the audio signal may further comprise setting a prescribed range for an intermittence level of the signal. Synthesising the proposed audio signal may further comprise setting the intermittence level between 1Hz and 2Hz. The above values, for forward driving and reverse driving respectively, have been found to provide particularly advantageous audio signals according to the direction of driving, where a greater level of attention is required for reverse driving compared to forward driving. In an example, the method may be an iterative method of designing the audio signal. The method may further comprise adjusting one or more sound characteristics of the synthesised audio signal for a subsequent iteration in dependence on the synthesised audio signal being determined invalid based on the comparison. According to another aspect of the invention, there is provided a computer program product, comprising computer readable instructions which, when the program is executed by one or more processors, cause the one or more processors to perform a method according to a previous aspect of the invention. According to yet another aspect of the invention, there is provided a computer readable medium comprising instructions which, when executed by a computer, cause the computer to perform the steps of a method as described in a previous aspect of the invention. According to a further aspect of the invention, there is provided an audio signal for an acoustic vehicle alerting system. The audio signal is designed by a method as described in a previous aspect of the invention. According to a still further aspect of the invention, there is provided an acoustic vehicle alerting system for a vehicle. The acoustic vehicle alerting system comprises an audio output device and a controller configured to output, via the audio output device, an audio signal as described in a previous aspect of the invention. According to yet another aspect of the invention, there is provided a vehicle comprising an acoustic vehicle alerting system as described in a previous aspect of the invention. Within the scope of this application it is expressly intended that the various aspects, embodiments, examples and alternatives set out in the preceding paragraphs, in the claims and / or in the following description and drawings, and in particular the individual features thereof, may be taken independently or in any combination. That is, all embodiments and / or features of any embodiment can be combined in any way and / or combination, unless such features are incompatible. The applicant reserves the right to change any originally filed claim or file any new claim accordingly, including the right to amend any originally filed claim to depend from and / or incorporate any feature of any other claim although not originally claimed in that manner. BRIEF DESCRIPTION OF THE DRAWINGS One or more embodiments of the invention will now be described, by way of example only, with reference to the accompanying drawings, in which: Figure 1 is a schematic illustration of an exemplary vehicle, including an acoustic vehicle alert system (AVAS), in accordance with an embodiment of the present invention; Figure 2 is a schematic illustration of the AVAS shown in Figure 1; Figure 3 schematically illustrates an exemplary method of designing an audio signal for the AVAS, shown in Figure 2, in accordance with an embodiment of the present invention; and Figure 4 schematically illustrates a comparison between first and second sound quality metrics, determined for a set of test signals according to the method of Figure 3, and respective metrics obtained by jury testing. DETAILED DESCRIPTION Embodiments of the present invention relate to an AVAS and to a method of designing an audio signal for the AVAS. As mentioned previously, various methods are known in the art for synthesizing audio signals for an AVAS and industry standards have been developed that recommend sound characteristics for alerting other road users to the presence of the vehicle. However, a wide range of signal variations exist within the recommendations and it has proven challenging to optimise the AVAS sound for listener responsiveness / perception. In this context, jury testing is a known methodology for comparing sounds in acoustics engineering and jury testing could feasibly be incorporated into the audio signal design process to provide feedback and improve listener perception. However, large numbers of participants are required to obtain meaningful data from jury testing, and the resulting design process is therefore challenging, expensive, and time-consuming to implement and repeat whenever a new signal is to be designed. In order to address these issues, the inventors have devised a method of designing an audio signal for an AVAS that includes an objective and empirical evaluation of listener response (without jury testing). In particular, synthesised audio signals are evaluated by determining first and second sound quality metrics for estimating respective metrics of listener response. The first and second sound quality metrics are compared to respective thresholds to test whether the listener response to the synthesised audio signal will be acceptable. The first sound quality metric is an annoyance level, which is indicative of relative annoyance experienced by listeners, and the second sound quality metric is a preference level, which is indicative of the relative ease of perception of the audio signal as a warning or relative listener alertness generated. The sound quality metrics themselves are determined for the synthesised audio signal based on a particular combination of psychoacoustic metrics. In particular, the first and second sound quality metrics are determined based on: a loudness, a sharpness, a fluctuation strength, and an articulation index, of the audio signal. In turn, each of these psychoacoustic metrics is determined based on measurable sounds characteristics of the synthesised audio signal. The first and second sound quality metrics have been found to strongly correlate with respective jury testing metrics and may advantageously be used to estimate the important metrics of listener response without the need for jury testing. In this manner, the first and second sound quality metrics are used as a proxy for evaluating the jury response in a method of designing an audio signal for an AVAS. In examples, the respective thresholds may therefore be derived from benchmarking (i.e., by analysing the results of existing AVAS audio signals, or other validated signals), and serves to test whether the synthesised audio signal will provide reduced annoyance and / or enhanced alertness. For example, the comparison may determine that the synthesised audio signal provides lower levels of annoyance whilst providing preference levels that may be consistent with industry standards. If the designed audio signal fails to satisfy the thresholds, the proposed audio signal may be rejected and / or reiterated, for example by adjusting one or more of the sound characteristics. In this manner, it is envisaged that embodiments of the invention will therefore provide an efficient and economical method of designing audio signals for an AVAS. The method allows for rapid development and evaluation to provide more effective and / or pleasant audio signals. It is envisaged that the audio signals designed according to embodiments of the invention will therefore have the effect of improving the listening experience of other road users and / or enhancing their alertness to the vehicle. Embodiments of the invention shall now be described in more detail with additional reference to Figures 1 to 4. Figure 1 shows a plan view of a vehicle 100, in accordance with an embodiment of the invention, including an electric powertrain 102, an Acoustic Vehicle Alert System (AVAS) 104, a sensor system 108, and an optional user interface device 110. In this example, the vehicle 100 takes the form of a fully electric passenger car. However, this example is not intended to be limiting on the scope of the invention, which is more generally applicable to quiet road transport vehicles. Accordingly, in other examples, the vehicle 100 may take various other forms, such as a bus, or truck, amongst other passenger vehicles, and the vehicle 100 may take the form of an electric vehicle ora hybrid electric vehicle, including both internal combustion and electric powertrain systems. The electric powertrain 102 include one or more electric machines or motors (not shown) for converting electrical energy from a battery, such as a high-voltage (HV) battery, on board the vehicle 100 into propulsive power. The invention is not intended to be limited to any specific configuration of the electric powertrain 102. However, it shall be appreciated that, in contrast to a conventional ICE vehicle, the electric powertrain generates much less noise during driving, particularly at low driving speeds and in urban driving settings. The AVAS 104 is therefore configured to emit sounds to alert pedestrians and other road users to the presence of the vehicle 100, accounting for the relatively low noise levels of the electric powertrain 104. The sounds emitted from the AVAS help other road users, especially those with visual impairments, to detect the vehicle 100 and recognise its proximity, for example. AVAS sounds are typically designed to comply with requirements established by respective industry standards and regulations. The emitted sounds are generated by outputting respective audio signals and may take various suitable forms for this purpose. For example, the emitted sounds may mimic engine sounds and / or tyre sounds, or otherwise include intermittent chirps or beeps for attracting attention, amongst other examples. As is typical for such vehicles, the AVAS 104 may be configured to emit such sounds in dependence on detecting one or more driving conditions of the vehicle 100. For this purpose, the AVAS 104 may be configured to receive indications of the driving conditions from one or more vehicle control systems (not shown) or to receive sensor signals, indicative of the driving conditions, from the sensor system 108. For example, the AVAS 104 may be configured to output an audio signal when the vehicle speed is less than a threshold vehicle speed and / or when the vehicle 100 is performing a particular manoeuvre, such as rearward driving (i.e. driving in a reverse direction). Moreover, in examples, the AVAS 104 may be configured to output different audio signals for respective driving manoeuvres, for example outputting a first audio signal when driving in a forward direction below a threshold speed and outputting a second audio signal when driving in a reverse direction. The first audio signal may, for example, mimic an engine sound, while the second audio signal may produce an intermittent beep. The sensor system 108 is therefore configured to monitor the vehicle operation and may include one or more sensors (not shown) for outputting respective sensor signals indicative of the current driving conditions of the vehicle 100. For example, such sensor(s) may include one or more vehicle speed sensors, steering sensors, and / or driving direction sensors, amongst others. The AVAS 104 may be configured to determine whether to output an audio signal and / or to select one of a plurality of audio signals to output in dependence on the received sensor signals and / or the indicated driving conditions. Such manners of operating an AVAS are well-known known in the art and are therefore not described in detail here to avoid obscuring the invention. The user interface device 110 is an optional system that may be included to allow a user to provide one or more inputs to the AVAS 104, for example to configure, or reconfigure, the sounds emitted by the AVAS 104. In an example, the user interface 110 may be configured to receive user inputs for selecting an audio signal for a respective driving condition, configuring the driving conditions for outputting that audio signal, and / or designing further audio signals according to the methods described herein. In some examples, the user interface 110 may include a display and one or more input devices, such as touchscreens, keypads, buttons and / or the like, which are configured to receive inputs from the user. In other embodiments, the user interface 110 may take the form of a touchscreen mounted on a dashboard of the vehicle 100. In still further embodiments, the user interface 110 may also be a smartphone, tablet, or other such user device or computing device, connected to the AVAS 104. An exemplary AVAS 104, in accordance with an embodiment of the invention, shall now be discussed with additional reference to Figure 2. In this example, the AVAS 104 is shown to include one or more controllers 114 and one or more audio output devices 116. However, it shall be appreciated that the controller(s) 114 and / or audio output device(s) 116 may form part of an independent system or part of other functional systems of the vehicle 100 (not limited to the AVAS). The controller(s) 114 are configured to receive inputs indicative of the driving conditions of the vehicle 100 and to determine corresponding audio signals for output to the audio output device(s) 116. For this purpose, the controller(s) 114 are shown to include an input module 118, an output module 120, a processor module 122, and a memory module 124, in this example. That is, the controller(s) 114 are shown to include four functional modules, units, or sub-systems in Figure 2. Each of these modules, units or sub-systems may be provided, at least in part, by suitable software running on any suitable computing substrate using conventional or custom processors and memory. Some or all of the units or modules may use a common computing substrate (for example, they may run on the same server) or separate substrates, or different combinations of the modules may be distributed between multiple computing devices. The example architecture of the controller 114 is not intending to be limiting on the scope of the invention though and, in other examples, it shall be appreciated that the architecture may take other suitable forms. The input module 118 is configured to receive signals indicative of the driving conditions of the vehicle 100. For example, such signals may include one or more signals received from a vehicle control system (not shown) monitoring the vehicle operation and / or one or more sensor signals received from the sensor system 108. For example, the input module 118 may be configured to receive sensor signals indicative of the vehicle speed and / or driving direction from the sensor system 108. In other examples, the input module 118 may be further configured to receive sensor signals indicative of the location of the vehicle 100 or the proximity to other road users, for example. The input module 118 may also be configured to receive inputs from a user, via the interface device 110. For example, the input module 118 may be configured to receive one or more inputs from the user interface device for configuring or selecting one of a plurality of audio signals to output in respective driving conditions. The processor module 122 is configured to determine a driving condition of the vehicle 100 based on the signals received from the input module 118, and to output an audio signal to the audio output device(s) 116 in dependence on the determined driving condition. For example, the processor module 122 may be configured to determine whether the vehicle speed indicated by a sensor signal is less than a threshold speed for outputting a first audio signal. In another example, the processor module 122 may additionally or alternatively be configured to determine whether the vehicle 100 is driving in a reverse driving direction based on the received sensor signals, and to output a second audio signal in dependence on detecting the reverse driving condition. These examples are not intended to be limiting on the scope of the invention though and it shall be appreciated that the processor module 122 may be configured to determine further driving conditions and respective audio signals. For this purpose, one or more driving conditions may be stored in the memory module 124 along with respective audio signals to output to the audio output device(s) 116. For example, such driving conditions and the respective audio signals may be stored in a lookup table of the memory module 124. The processor module 122 may be configured to access the lookup table and to compare the driving conditions indicated by the received signals to the lookup table in order to determine a corresponding audio signal to output. The output module 120 is configured to output the determined audio signal to the audio output device(s) 116, and thereby control the audio output device(s) 116 to emit a corresponding AVAS sound. The audio output device(s) 116 may include one or more speakers or actuators (not shown) for emitting AVAS sounds externally to other road users. The audio output device(s) 116 may, for example, include a controller for operating the speaker(s) and / or actuator(s) to vibrate the structure of the vehicle 100 and produce a corresponding sound. To give an example, the audio output device(s) 116 may include a single speaker on a front bumper of the vehicle 100 for emitting the AVAS sounds. In other examples, the audio output device(s) 116 may include a plurality of speakers dispersed around a body of the vehicle 100 for creating the AVAS sounds. In such examples, beamforming techniques may be used to create directed AVAS sounds using the plurality of speakers. The AVAS 104 is therefore configured to receive signal(s) indicative of the driving condition, determine the driving condition, and emit a respective AVAS sound for alerting other road users to the presence of the vehicle 100 in dependence thereon. Advantageously, each AVAS sound is designed according to methods of the present invention to mitigate listener annoyance metrics and increase listener preference metrics relative to conventional AVAS sounds. An exemplary method, in accordance with embodiments of the present invention, of designing an audio signal for the AVAS 104 is described below with additional reference to Figure 3. Figure 3 shows an exemplary method 200 of designing an audio signal for the AVAS 104. The method 200 itself is computer-implemented and may be executed as a computer program or using software running on any suitable computing substrate, or combination of computing substrates, for example with conventional or custom processors and memory. The method 200 may therefore be executed by a manufacturer of the vehicle 100, the AVAS 104, or parts thereof. Alternatively or additionally, the method 200 may be executed by a user, via the user interface device 110, to design or adjust audio signals, for example to suit individual user preferences. In step 202, a proposed audio signal is synthesized for the AVAS 104 using one or more synthesis methods that are known in the art. To give an example, the audio signal may be synthesised using an order-based synthesis method configured to synthesise an audio signal that mimics engine noise. In other examples, the proposed audio signal may be synthesised by a wavetable synthesis method and / or by a granular synthesis method. The wavetable synthesis method involves periodic reproduction of multiple single-cycle waveforms with digital interpolation between adjacent waveforms, while the granular synthesis method splits an original sample sound into segments, or grains, that are reorganised according to the method to synthesise the audio signal. In each case, the synthesis method, used in step 202, is configured to ensure that a set of sounds characteristics of the synthesised audio signal fall within respective ranges prescribed for each sound characteristic. By way of example, such sound characteristics may include one or more characteristics of a frequency spectrum of the proposed audio signal, an amplitude modulation of the signal, a delta of the amplitude modulation, and / or an intermittence level or modulation frequency of the signal, amongst other sounds characteristics. The prescribed ranges may correspond to local or international standards and regulations for the design and certification of audible emergency signals, for example, amongst other factors. For example, in the United Kingdom, the prescribed ranges may comply with industry standards, such as ISO 7731:2018 and EN61310:2088, which define basic physical principles of an effective audio signal for an AVAS. It shall be appreciated that the prescribed ranges may therefore be pre-programmed parameters of the synthesis method. To give an example, an energy / sound pressure level (SPL) concentration of the frequency spectrum may be limited to between 500 and 2500 Hz; an amplitude modulation of the signal may be set to less than or equal to 30dBA; and / or an an intermittence of the signal may be limited to between 0.5 and 4 Hz. In an example, frequencies greater than 1200Hz may be avoided entirely in the frequency spectrum, i.e. with a zero or negligible energy / sound pressure level (SPL) concentration. The synthesis method may therefore be provided with such constraints and generate a compliant audio signal, in step 202. Through extensive testing, the inventors have also determined preferred ranges for respective sound characteristics that lead to audio signal improvements. In particular, the listener response to a plurality of audio signals can be evaluated, according to the methods described herein, some of the audio signals having been synthesised for this purpose and others having been taken from existing AVAS systems. The listener responses to these audio signals can then be used to rank the signals and analyse the sounds characteristics that lead to improvements of the listener response. By virtue of this analysis, preferred ranges for limiting the respective sound characteristics can be identified, where such preferred ranges are correlated with improvements in the listener response metrics. In examples, the synthesis method, used in step 202, may therefore be configured to ensure that the respective sounds characteristics of the synthesised audio signal fall within the preferred ranges and / or overlapping sub-range of the prescribed and preferred ranges. By way of example, the preferred ranges may include: an energy / sound pressure level (SPL) concentration of the frequency spectrum between 200 and 1000 Hz; an amplitude modulation of the signal of less than or equal to 10dBA; and / or an an intermittence of the signal may be limited to between 1 and 2 Hz. Accordingly, the overlapping sub-ranges incorporated into the synthesis method, used in step 202, may include: an energy / sound pressure level (SPL) concentration of the frequency spectrum between 500 and 1000 Hz; an amplitude modulation of the signal of less than or equal to 10dBA; and / or an intermittence of the signal may be limited to between 1 and 2 Hz. In an example, frequencies greater than 1200Hz may be excluded from the frequency spectrum, i.e. with a zero or negligible energy / sound pressure level (SPL) concentration. It shall be appreciated that different ranges for the sound characteristics may also be applicable to respective driving conditions. For example, the ranges for a particular sound characteristic may vary with a particular driving speed and / or driving direction. The ranges used for the respective sound characteristics in the synthesis methods may therefore be associated with respective driving conditions, and thereby synthesise audio signals for respective driving conditions, in step 202. For example, when synthesising an audio signal for a forward driving condition, the intermittence of the signal may be limited to 0 Hz, while the intermittence may be limited to any of the ranges described above for a reverse driving condition. In this manner, the synthesis method ensures that the audio signal is intermittent for reverse driving purposes, increasing the listener perception of said signal as an alert or alarm, whilst there is no intermittence in the audio signal for forward driving purposes to minimise listener discomfort or annoyance. Similarly, when synthesising an audio signal for a forward driving condition, the amplitude modulation of the signal may be limited to between 3 to 6 dBA, while the amplitude modulation may be limited to the ranges described above for a reverse driving condition. In this manner, the method 200 is able to produce different audio signals for respective driving conditions, accounting for the varying expectations or requirements of such driving conditions. In step 204, a set of psychoacoustic metrics are determined based on the synthesised audio signal. In particular, the synthesised audio signal is processed to determine: a loudness; a sharpness; a fluctuation strength; and an articulation index of the signal. In this context, the ‘Loudness’ refers to a subjective perception of sound pressure, which is measured on a sone scale to further account for the human hearing domain (which is both frequency and level dependent). The ‘Sharpness’ is a ratio of a weighted loudness to full bandwidth loudness and provides an objective way of quantifying the amount of high frequency content present in the audio signal. The sharpness is measured in units of “acum”, where 1 acum corresponds to the sharpness of a broad-band noise centred on 1 kHz, with a width of 1 critical band and a level of 60 dB. The ‘Fluctuation strength’ quantifies the amount of modulation present in a sound and is used to describe sounds with 20 modulations per second or less. The unit of measure for fluctuation strength is the vacil. One vacil is defined as the fluctuation strength produced by a 1000Hz tone of 60dB which is 100% amplitude modulated at 4Hz. The ‘Articulation Index’ (Al) is a sound metric that indicates how much the sound levels interfere with human speech. The Al has a value that ranges between 0% (no speech understood) to 100% (all speech understood) and is influenced by human speech frequency range, as well as masking and octave bands. Methods for determining each of the psychoacoustic metrics are well known in the art and are not described in detail here to avoid obscuring the invention. However, to give an example, whilst there are different standards available for the calculation of a loudness level (sones), the most commonly used methods are detailed in the international standard ISO 532B Acoustics - Part 1: Zwicker method and in the German Standard DIN 45631. According to such methods, the loudness level is obtained based on human perception by considering the frequency content, the sound pressure level, the directional response pattern (free / diffuse field), the auditory masking, and temporal masking, among other parameters of the sound signal. With increasing loudness, sounds attract more attention, as they are perceived to have a higher volume, without showing a difference in an A-weighted sound pressure level. The method used to calculate the sharpness may be based on a prior loudness calculation according to the Zwicker method, as described in DIN 45692 for example. The sharpness may be computed as a weighted sum of the specific loudness levels reported to the global loudness. Several weighting functions exist (by Aures, Fasti, Von Bismarck), and they all typically increase toward the highest frequency bands to model the fact that the human hearing system is more sensitive to the levels of high-frequency components. As sharpness increases, sounds may be perceived as increasingly aggressive and / or annoying without showing a difference in volume or A-weighted sound pressure level. The method used to calculate Fluctuation strength may be based on the International Standard ECMA 418-2, for example, which describes the methods used to quantify the modulation frequency and the modulation level of a sound signal. Typically, fluctuation Strength describes modulation at frequencies below 20 Hz where, because the sound varies slowly over time, a listener can hear each individual rise and fall in the sound. With increasing fluctuation strength, sounds may attract more attention and be perceived as increasingly aggressive, and annoying, without showing a difference in loudness or A-weighted sound pressure level. The method defined in ANSI S3.5-1969 and ANSI S3.5-1997, may be used for calculating the ‘Articulation Index’. ANSI S3.5-1969 and ANSI S3.5-1997 define a method for computing a physical measure that is highly correlated with the intelligibility of speech, as evaluated by speech perception tests, for a given group of talkers and listeners. As the articulation rate increases, sounds may become increasingly understandable and clear to the human ear. While various psychoacoustic metrics are known in the art, the inventors have found that the above-mentioned set of psychoacoustic metrics are readily combined into first and second sound quality metrics for reliably assessing listener annoyance and preference metrics relating to the emitted sound, as shall be described in more detail. In step 206, the method 200 involves calculating first and second sound quality metrics for the synthesised audio signal based on the determined psychoacoustic metrics. The first sound quality metric is an annoyance level, which is indicative of relative annoyance scores reported by listeners in response to the audio signal. The second sound quality metric is a preference level, which is indicative of relative scores rating the ease of perception of the audio signal as an audible warning or alarm. The annoyance level is calculated based on the determined sharpness, loudness, and articulation index. In this respect, the inventors have found that the sharpness and loudness of the synthesised audio signal correlate strongly with the reported annoyance scores, whilst the articulation index of the proposed audio signal has been found to share a negative or inverse correlation with the annoyance scores. In examples, the annoyance level is therefore determined as a rational function of at least the determined sharpness and the determined loudness, divided by the determined articulation index. In particular, as the loudness was found to have a particularly strong correlation with the reported annoyance scores, the loudness is raised to the power of 2 in examples to determine more accurate estimates. In examples, the annoyance level may therefore be determined according to Equation 1 below: Equation 1: . (AxSharpness}x(BxLoudness2} Annoyance = 21--------—---------- (CxArticulation Index) where A, B, and C are numerical constants, that may, for example, be equal to 1 or otherwise determined based on the correlation coefficients of the respective variables with reported annoyance scores obtained by separate jury testing of sample signals. Moreover, the inventors have found that the maximum value of the determined sharpness and the N10 value of the determined loudness correlate even more strongly with the reported annoyance scores. Here it shall be appreciated that N10 is the sone value that 10% of the loudness values are above for the duration of the audio signal. Accordingly, in examples, the annoyance level may be determined according to Equation 2 below: Equation 2: Annoyance = J (Articulation Index) The invention is not intended to be limited to this example though and, in other examples, a k-th percentile of the determined loudness of the proposed audio signal may be preferred, where k is a positive integer greater than or equal to 75, optionally, being greater than or equal to 90. The preference level is calculated based on the determined loudness and the determined fluctuation strength of the proposed audio signal. In this respect, the inventors have found that the loudness and fluctuation strength of the proposed audio signal correlate strongly with the reported preference scores. In examples, the preference level is therefore determined as a polynomial function of at least the determined loudness and fluctuation strength. For example, the preference level may be determined according to Equation 3 below: Equation 3: Preference = (D x Loudness') + (E x Fluctuation strength) — F, where D, E, and F are numerical constants and D and E may, for example, be determined based on the correlation coefficient of the respective variables with the reported preference scores in separate jury testing of sample signals. Through further testing, the inventors have found that the indicated preference level correlates most strongly with the jury testing results when the coefficient D is 0.190, the coefficient E is 0.409, and the coefficient F is 1.885. In examples, the annoyance and preference levels are therefore determined, in step 206, for the synthesised audio signal according to the respective equations described above. It shall be appreciated that the annoyance and preference levels determined according to such equations correlate with respective jury testing results with sufficiently strength that jury testing of the synthesised audio signal is not required. By way of example, Figure 4 shows the correspondence between jury testing results for a set of test signals 402, 404, 406, 408 and corresponding annoyance and preference levels, calculated according to Equations 2 and 3 respectively. In this instance, the jury testing was performed for a selection of four test signals 402, 404, 406, 408 used by existing AVASs. The jury was asked to rate the annoyance experienced and the ease of precepting each test signal as an alert. In particular, each participant was asked to rate the annoyance level they experienced whilst listening to each test signal on a scale of 1 to 5, with 0 being most annoying and 5 being most comfortable. The jury was further asked to rate the ease of recognising each test signal as an alert on a scale from 0 to 5, with 0 indicating the most difficult signal to recognise, and 5 being the most preferred and easily recognisable signal. As shown in Figure 4, a strong correlation was therefore produced between the subjective jury testing results and the first and second sound quality metrics for each test signal 402, 404, 406, 408. The objective analysis using the sound quality metrics has therefore been found through this, and further testing, to reliably predict listener responses. The sound quality metrics can therefore be used as a substitute for jury testing each signal synthesised in the method 200. In step 208, the calculated first and second sound quality metrics are compared to respective thresholds for validating the synthesised audio signal. The thresholds are used in this manner to test whether the listener response to the synthesised audio signal will be acceptable, for example testing whether the audio signal serves to adequately attract attention to the vehicle 100 during driving and / or sufficiently mitigate the resulting annoyance or discomfort of other road users. By way of example, when the preference level is determined according to Equation 3 above, the threshold used for the preference level may be greater than or equal to 0.25. When the annoyance level is determined according to Equation 2 above, the threshold used for the annoyance level may be less than or equal to 1. That is, the inventors of the present invention have found that a preference level threshold of less than or equal to 0.25 and an annoyance level threshold of less than or equal to 1, is particularly effective for ensuring high performing audio signals. Moreover, in examples, the applicable thresholds may depend on the respective driving conditions for outputting that audio signal. For example, the thresholds may depend on a particular driving speed and / or driving direction associated with the synthesised audio signal. In this manner, the annoyance level threshold may be reduced for forward driving compared to reverse driving, accounting for the relatively high risk to other road users during reverse driving and the corresponding desire for greater alertness to the vehicle 100. By way of example, it has been found to be particularly advantageous for the threshold for the annoyance level to be reduced from 1 (for reverse driving) in the above example to 0.75 (for forward driving). More generally, the thresholds for each of the annoyance and preference levels may be determined based on benchmarking against a set of test signals. For example, the set of test signals may include one or more previously synthesised audio signals, and / or one or more known audio signals obtained from vehicles available on the market. In each case, the test signals may be analysed in accordance with steps 204 and 206 of the method 200 to calculate respective annoyance and preference levels for each test signal. The respective thresholds, used in step 208, for the annoyance and preference levels may therefore be determined based on the preference and annoyance levels calculated for the test signals. To give an example, the threshold preference level may be determined based on a maximum value of the preference levels determined for the test signals. For example, the threshold preference level may be greater than or equal to the maximum preference level determined for the test signals. In this manner, the threshold may be set accordingly to test whether listeners will prefer the validated audio signal over the most preferred test signal. Likewise, the threshold annoyance level may be determined based on a minimum value of the annoyance levels determined for the test signals. For example, the threshold annoyance level may be less than or equal to the minimum annoyance level determined for the test signals. In this manner, the threshold may be set accordingly to test whether listeners will find the validated audio signal less annoying than the least annoying test signal. In step 210, it is therefore determined whether the first and second sound quality metrics of the synthesised audio signal satisfy the respective thresholds for validation. If one or both of the sound quality metrics do not satisfy the respective thresholds, then the synthesised audio signal is rejected. Optionally, following rejection, the sound characteristics may be adjusted within the prescribed ranges to refine the audio signal, as shown in step 214 of Figure 3. The refined audio signal may then be re-evaluated, in a subsequent iteration, according to steps 204 to 210. Alternatively, if the first and second sound quality metrics are determined to satisfy the respective thresholds, the method 200 proceeds to validate the audio signal for use in the AVAS 104, in step 212. The validated audio signal can subsequently be output by the AVAS 104 to effectively alert other road users to the presence of the vehicle 100, in a pleasant and comfortable manner. Validated audio signals may therefore be recorded in the memory module 124 of the AVAS 104 for subsequent recall. For example, the valid audio signals may be stored in the memory module 124 of the AVAS 104 with a respective driving condition for subsequent recall during driving of the vehicle 100, substantially as described previously. The method 200 may therefore be repeated to determine a set of validated audio signals for respective driving conditions, as used by the AVAS 104 during driving, this has already been described above through the adjusting of the annoyance level threshold value for forward vs reverse driving directions. As an additional example, different audio signals may be used in dependence on the speed at which the vehicle 100 is moving. In one embodiment, one audio signal may be determined for use when the vehicle 100 is travelling at speeds less than or equal to 10 mph and another audio signal may be determined for use when the vehicle 100 is travelling at speeds greater than 10mph. This may be particularly advantageous in that the vehicle 100 is more likely to be travelling at low speeds in areas in which it is more common to find pedestrians, e.g. a car park, and the vehicle 100 is more likely to be travelling at high speeds in areas in which it is less common to find pedestrians, e.g. a main road. It shall be appreciated that the output audio signals have therefore been evaluated and found to provide acceptable listener response without need for conducting a jury test of each signal. The method 200 therefore provides a simple, low-cost, and time efficient means of designing approved audio signals for an AVAS. As a result, the AVAS 104 is configured to output audio signals that produce enhanced alertness to the vehicle 100 and improved listener perception. It shall be appreciated that various changes and modifications can be made to the examples described above without departing from the scope of the present invention. For example, it shall be appreciated that it is also possible to validate existing audio 5 signal for an AVAS according to steps 202 to 212, for example to retrospectively evaluate an audio signal designed according to another method

Claims

1. A method of designing an audio signal for an acoustic vehicle alerting system, the method comprising:synthesising an audio signal for the acoustic vehicle alerting system using one or more synthesis methods;determining a set of psychoacoustic metrics based on the synthesised audio signal, the set of psychoacoustic metrics comprising:a loudness;a sharpness;a fluctuation strength; andan articulation index;calculating first and second sound quality metrics based on the determined set of psychoacoustic metrics,the first sound quality metric being an annoyance level of the proposed audio signal calculated based on a rational function of at least the determined sharpness and loudness divided by the determined articulation index,the second sound quality metric being a preference level calculated based on a polynomial function of at least the determined loudness and fluctuation strength;comparing the first and second sound quality metrics to respective thresholds; andvalidating the proposed audio signal for the acoustic vehicle alerting system in dependence on the comparison to the thresholds.

2. A method according to claim 1, wherein the rational function takes the form of:. (AxSharpness}x(BxLoudness2}Annoyance = ---------—---------(Cx Articulation Index)where A, B, and C are numerical constants and, optionally, A, B, and C are each equal to 1.

3. A method according to claim 2, wherein the annoyance level is determined based on:a maximum value of the determined sharpness of the proposed audio signal; and / or a k-th percentile of the determined loudness of the proposed audio signal, where k is a positive integer greater than or equal to 75 and, optionally, k is a positive integer greater than or equal to 90.

4. A method according to any preceding claim, wherein the rational function is determined based on a regression analysis of:listener annoyance values indicated by jury test data obtained for a set of test signals, each test signal being a respective audio signal for an acoustic vehicle alerting system; andthe sharpness, loudness, and articulation index determined for each test signal.

5. A method according to any preceding claim, wherein the linear polynomial function takes the form of:Preference = (D x Loudness) + (E x Fluctuation strength) — F,where D, E, and F are numerical constants and optionally, D is equal to 0.19, E is equal to 0.409 and F is equal to 1.885.

6. A method according to any preceding claim, wherein the polynomial function is determined based on a regression analysis of:listener preference levels indicated by jury test data obtained for a set of test signals, each test signal being a respective audio signal for an acoustic vehicle alerting system; and the loudness and fluctuation strength determined for each test signal.

7. A method according to any preceding claim, wherein synthesising the audio signal comprises setting a prescribed range for each of the following sound characteristics:a frequency spectrum;a sound pressure level; andan amplitude modulation.

8. A method according to claim 7, wherein synthesising the proposed audio signal comprises:setting the frequency spectrum of the audio signal to include at least 65% of the energy and / or sound pressure level (SPL) at frequencies between 500Hz and 1000Hz and, optionally, at least 90% of the energy and / or sound pressure level (SPL) at frequencies between 500Hz and 1000Hz; andsetting the amplitude modulation of the audio signal less than or equal to 10dBA.

9. A method according to claim 8, wherein the audio signal is designed for a forward driving condition, and wherein the amplitude modulation of the audio signal is set between 3dBA and 6dBA.

10. A method according to claim 7 or claim 8, wherein the audio signal is designed for a reverse driving condition, and wherein synthesising the audio signal further comprises setting a prescribed range for an intermittence level of the signal.

11. A method according to claim 10, when dependent on claim 8, wherein synthesising the proposed audio signal comprises:setting the intermittence level between 1 Hz and 2Hz.

12. A method according to any preceding claim, wherein the method is an iterative method of designing the audio signal further comprising adjusting one or more sound characteristics of the synthesised audio signal for a subsequent iteration in dependence on the synthesised audio signal being determined invalid based on the comparison.

13. A computer program product, comprising computer readable instructions which, when the program is executed by one or more processors, cause the one or more processors to perform a method according to any preceding claim.

14. A computer readable medium comprising instructions which, when executed by a computer, cause the computer to perform the steps of a method according to any of claims 1 to 12.5 15. An acoustic vehicle alerting system for a vehicle, the acoustic vehiclealerting system comprising an audio output device and a controller configured to output, via the audio output device, an audio signal designed by a method according to any of claims 1 to 12.10s