Facilitation of speech recognition in user interface6912500Abstract Items are represented to a user through a user interface with each item having a respective perceivable range value and associated label by which the item can be addressed. To address a particular item, the user speaks its label at a loudness indicative of its perceived range. A loudness-to-range function of the interface determines on the basis of the loudness of the user input, a range gate expected to encompass the range value of the addressed item. A speech recogniser is used to recognise the spoken label and thus the addressed item, the label search space of the recogniser being restricted to exclude the labels of items having a range value outside of the determined range gate. In one embodiment, the user interface is an audio interface in which the items are represented in an audio field through corresponding synthesized sound sources, the depth at which each sound source is rendered in the audio field being the range value associated with the corresponding item. Claims 1. A user-interface method in which items are represented to a user with respective perceivable range values, the items having respective associated labels by which they can be addressed, the method involving: Description FIELD OF THE INVENTION
It may also be noted that in order avoid source-localization errors arising from sound reflections, humans localize sound sources on the basis of sounds that reach the ears first (an exception is where the direct/reverberant ratio is used for range determination). Getting a sound system (sound producing apparatus) to output sounds that will be localized by a hearer to desired locations, is not a straight-forward task and generally requires an understanding of the foregoing cues. Simple stereo sound systems with left and right speakers or headphones can readily simulate sound sources at different azimuth positions; however, adding variations in range and elevation is much more complex. One known approach to producing a 3D audio field that is often used in cinemas and theatres, is to use many loudspeakers situated around the listener (in practice, it is possible to use one large speaker for the low frequency content and many small speakers for the high-frequency content, as the auditory system will tend to localize on the basis of the high frequency component, this effect being known as the Franssen effect). Such many-speaker systems are not, however, practical for most situations. For sound sources that have a fixed presentation (non-interactive), it is possible to produce convincing 3D audio through headphones simply by recording the sounds that would be heard at left and right eardrums were the hearer actually present. Such recordings, known as binaural recordings, have certain disadvantages including the need for headphones, the lack of interactive controllability of the source location, and unreliable elevation effects due to the variation in pinna shapes between different hearers. To enable a sound source to be variably positioned in a 3D audio field, a number of systems have evolved that are based on a transfer function relating source sound pressures to ear drum sound pressures. This transfer function is known as the Head Related Transfer Function (HRTF) and the associated impulse response, as the Head Related Impulse Response (HRIR). If the HRTF is known for the left and right ears, binaural signals can be synthesized from a monaural source. By storing measured HRTF (or HRIR) values for various source locations, the location of a source can be interactively varied simply by choosing and applying the appropriate stored values to the sound source to produce left and right channel outputs. A number of commercial 3D audio systems exist utilizing this principle. Rather than storing values, the HRTF can be modeled but this requires considerably more processing power. The generation of binaural signals as described above is directly applicable to headphone systems. However, the situation is more complex where stereo loudspeakers are used for sound output because sound from both speakers can reach both ears. In one solution, the transfer functions between each speaker and each ear are additionally derived and used to try to cancel out cross-talk from the left speaker to the right ear and from the right speaker to the left ear. Other approaches to those outlined above for the generation of 3D audio fields are also possible as will be appreciated by persons skilled in the art. Regardless of the method of generation of the audio field, most 3D audio systems are, in practice, generally effective in achieving azimuth positioning but less effective for elevation and range. However, in many applications this is not a particular problem since azimuth positioning is normally the most important. As a result, systems for the generation of audio fields giving the perception of physically separated sound sources range from full 3D systems, through two dimensional systems (giving, for example, azimuth and elevation position variation), to one-dimensional systems typically giving only azimuth position variation (such as a standard stereo sound system). Clearly, 2D and particularly 1D systems are technically less complex than 3D systems as illustrated by the fact that stereo sound systems have been around for very many years. In terms of user experience, headphone-based systems are inherently "head stabilized"-that is, the generated audio field rotates with the head and thus the position of each sound source appears stable with respect to the user's head. In contrast, loudspeaker-based systems are inherently "world stabilized" with the generated audio field remaining fixed as the user rotates their head, each sound source appearing to keep its absolute position when the hearer's head is turned. In fact, it is possible to make headphone-based systems "world stabilized" or loudspeaker-based systems "head stabilized" by using head-tracker apparatus to sense head rotation relative to a fixed frame of reference and feed corresponding signals to the audio field generation system, these signals being used to modify the sound source positions to achieve the desired effect. A third type of stabilization is also sometimes used in which the audio field rotates with the user's body rather than with their head so that a user can vary the perceived positions of the sound sources by rotating their head; such "body stabilized" systems can be achieved, for example, by using a loudspeaker-based system with small loudspeakers mounted on the user's upper body or by a headphone-based system used in conjunction with head tracker apparatus sensing head rotation relative to the user's body. As regards the purpose of the generated audio field, this is frequently used to provide a complete user experience either alone or in conjunction with other artificially-generated sensory inputs. For example, the audio field may be associated with a computer game or other artificial environment of varying degree of user immersion (including total sensory immersion). As another example, the audio field may be generated by an audio browser operative to represent page structure by spatial location. Alternatively, the audio field may be used to supplement a user's real world experience by providing sound cues and information relevant to the user's current real-world situation. In this context, the audio field is providing a level of "augmented reality". It is an object of the present invention to facilitate speech recognition in user interfaces. SUMMARY OF THE INVENTION According to one aspect of the present invention, there is provided a user-interface method in which items are represented to a user with respective perceivable range values, the items having respective associated labels by which they can be addressed, the method involving: According to another aspect of the present invention, there is provided user-interface apparatus comprising:
BRIEF DESCRIPTION OF THE DRAWINGS Embodiments of the invention will now be described, by way of non-limiting example, with reference to the accompanying diagrammatic drawings, in which: FIG. 1 is a functional block diagram of a first audio-field generating apparatus; FIG. 2 is a diagram illustrating a coordinate system for positions in a spherical audio field; FIG. 3 is a diagram illustrating rotation of an audio field relative to a presentation reference vector; FIG. 4 is a diagram illustrating a user exploring a body-stabilized audio field by head rotation; FIG. 5 is a diagram illustrating a user exploring a body-stabilized audio field by rotating the field in azimuth; FIG. 6 is a diagram illustrating a general cylindrical organization of an audio field; FIG. 7 is a diagram illustrating a first specific form of the FIG. 6 cylindrical organization; FIG. 8 is a diagram illustrating a second specific form of the FIG. 6 cylindrical organization; FIG. 9 is a functional block diagram of a variant of the FIG. 1 apparatus; FIG. 10 is a functional block diagram of a second audio-field generating apparatus; FIG. 11 is a diagram illustrating the operation of a focus expander of the FIG. 10 apparatus to expand an audio field, the user facing in the same direction as an audio field reference vector; FIG. 12 is a further diagram illustrating the operation of the focus expander, the user in this case facing in a different direction to the audio field reference vector; FIG. 13 is a diagram illustrating the operation of a segment muting filter of the FIG. 10 apparatus; FIG. 14 is a diagram illustrating the operation of a cyclic muting filter of the FIG. 10 apparatus; FIG. 15 is a diagram illustrating the operation of a collection collapser of the FIG. 10 apparatus; FIG. 16 is a diagram illustrating the operation of a range sound setter of the FIG. 10 apparatus; FIG. 17 is a diagram illustrating the concept of the range sound setter applied to a context of a fixed device being approached by a person; FIG. 18 is a functional block diagram showing further detail of the FIG. 10 apparatus; FIG. 19 is a diagram showing a relationship between loudness of a speech input and a range gate set by the FIG. 10 apparatus for limiting the search space of a speech recognizer of the apparatus; FIG. 20 is a diagram of a trackball type of input device usable by the FIG. 10 apparatus; FIG. 21 is a diagram showing a trackball input device similar to FIG. 20 but including a first form of visual orientation indicator arrangement; FIG. 22 is a block diagram of functionality for determining the orientation of the audio field relative to an indicator reference; FIG. 23 is a diagram showing a trackball input device similar to FIG. 20 but including a second form of visual orientation indicator arrangement; and FIG. 24 is a diagram of another form of input device usable by the FIG. 10 apparatus, this device being suitable where the apparatus is arranged to produce a cylindrical audio field; and BEST MODE OF CARRYING OUT THE INVENTION The forms of apparatus to be described below are operative to produce an audio field to serve as an audio interface to services such as communication services (for example, e-mail, voice mail, fax, telephone, etc.), entertainment services (such as internet radio), information resources (including databases, search engines and individual documents), transactional services (for example, retail and banking web sites), augmented-reality services, etc. When the apparatus is in a "desktop" mode, each service is represented in the audio field through a corresponding synthesized sound source presenting an audio label (or "earcon") for the service. The audio label associated with a service can be constituted by any convenient audio element suitable for identifying that service-for example, an audio label can be the service name, a short verbal descriptor, a characteristic sound or jingle, or even a low-level audio feed from the service itself. The sound sources representing the services are synthesized to sound, to a user, as though they exist at respective locations in the audio field using any appropriate spatialisation method; these sound sources do not individually exist as physical sound output devices though, of course, such devices are involved in the process of synthesizing the sound sources. Furthermore, the sound sources only have a real-world existence to the extent that service-related sounds are presented at the sound-source locations. Nevertheless, the concept of sound sources located at specific locations in the audio field is useful as it enables the sound content that is to be presented in respect of a service to be disassociated from the location and other presentation parameters for those sounds, these parameters being treated as associated with the corresponding sound source. Thus, the present specification is written in terms of such sound sources spatialized to specific locations in the audio field. Upon a service presented through a sound source being selected (in a manner to be described hereinafter), the apparatus changes from the desktop mode to a service mode in which only the selected service is output, a fall service audio feed now being presented in whatever sound spatialisation is appropriate for the service. When a user has finished using the selected service, the user can switch back to the desktop mode. It will be appreciated that other possibilities exist as to how the services are presented and accessed-for example, the feed from a selected service can be output simultaneously with background presentation of audio labels for the other available services. Furthermore, a service can provide its data in any form capable of being converted in audible form; for example, a service may provide its audio label in text form for conversion by a text-to-speech converter into audio signals, and its full service feed as digitised audio waveform signals. It is also possible in the desktop mode to use more than one sound source to represent a particular service and/or to associate more than one audio label with each sound source as will be seen hereinafter. Audio Field Organisation-Spherical Field Example Considering now the first apparatus (FIG. 1), in the form of the apparatus primarily to be described below, the audio field is a 2D audio field configured as the surface of a sphere (or part of a sphere). Such a spherical-surface audio field is depicted in FIG. 2 where a spatialised sound source 40 (that is, a service audio label that has been generated so as to appear to come from a particular location in the audio field) is represented as a hexagon positioned on the surface of a sphere 41 (illustrated in dashed outline). It maybe noted that although such a spherical surface exists in three-dimensional space, the audio field is considered to be a 2 dimensional field because the position of spatialised sound sources in the audio field, such as source 40, can be specified by two orthogonal measures; in the present case these measures are an azimuth angle X° and an elevation angle Y°. The azimuth angle is measured relative to an audio-field reference vector 42 that lies in a horizontal plane 43 and extends from the centre of sphere 41. The elevation angle is the angle between the horizontal and the line joining the centre of the sphere and the sound source 40. In fact, the FIG. 1 apparatus is readily adapted to generate a 3D audio field with the third dimension being a range measure Z, also depicted in FIG. 2, that is the distance from the centre of sphere 41 to the spatialised sound source 40. Conversely, the FIG. 1 apparatus can be adapted to generate a 1D audio field by doing away with the elevation dimension of the spatialised sound sources. The FIG. 1 apparatus supports azimuth rotation of the audio field, this potentially being required for implementing a particular stabilization (that is, for example, head, body, vehicle or world stabilization) of the audio field as well as providing a way for the user to explore the audio field by commanding a particular rotation of the audio field. As is illustrated in FIG. 3, the azimuth rotation of the field can be expressed in terms of the angle R between the audio-field reference vector 42 and a presentation reference vector 44. This presentation reference vector corresponds to the straight-ahead centreline direction for the configuration of audio output devices 11 being used. Thus, for a pair of fixed, spaced loudspeakers, the presentation reference vector 44 is the line of equidistance from both speakers and is therefore itself fixed relative to the world; for a set of headphones, the presentation reference vector 44 is the forward facing direction of the user and therefore changes its direction as the user turns their head. When the field rotation angle R=0°, the audio-field reference vector 42 is aligned with the presentation reference vector 44. The user is at least notionally located at the origin of the presentation reference vector. The actual position at which a service-representing sound source is to be rendered in the audio output field (its "rendering position") by the FIG. 1 apparatus, must be derived relative to the presentation reference vector since this is the reference used by the spatialisation processor 10 of the apparatus. The rendering position of a sound source is a combination of the intended position of the source in the audio field judged relative to the audio-field reference vector, and the current rotation of the audio field reference vector relative to the presentation reference vector. As already intimated, apart from any specific azimuth rotation of the audio field deliberately set by the user, the audio field may need to be rotated in azimuth to provide a particular audio-field stabilisation. Whether this is required depends on the selected audio-field stabilization and the form of audio output devices. Thus, by way of example, unless otherwise stated, it will be assumed below that the audio output devices 11 of FIG. 1 apparatus are headphones and the audio field is to be body-stabilised so that the orientation of the audio field relative to the user's body is unaltered when the user turns their head-this is achieved by rotation of the audio field relative to the presentation reference vector for which purpose a suitable head-tracker sensor 33 is provided to measure the azimuth rotation of the user's head relative to its straight-ahead position (that is, relative to the user's body). As the user turns their head, the angle measured by sensor 33 is used to rotate the audio field by the same amount but in the opposite direction thereby stabilising the rendering positions of the sound sources relative to the user's body. It will be appreciated that had it been decided to head-stabilise the field, then for audio output devices in the form of headphones, it would have been unnecessary to modify the orientation of the audio field as the user turned their head and, in this case, there would be no need for the head-tracker sensor 33. This would also be true had the audio output devices 11 taken the form of fixed loudspeakers and the audio field was to be world-stabilized. Where headphones are to be used and the audio field is to be world stabilised, the orientation of the audio field must be modified by any change in orientation of the user's head relative to the world, whether caused by the user turning their head or by body movements; a suitable head-tracker can be provided by a head-mounted electronic compass. Similarly, if the audio output devices 11 are to be provided by a vehicle sound system and the audio field is to be world stabilised, the orientation of the audio field must be modified by any change in orientation of the vehicle as determined by any suitable sensor. It may be generally be noted that where a user is travelling in a vehicle, the latter serves as a local world so that providing vehicle stabilisation of the audio field is akin to providing world stabilisation (whether the audio output devices are headphones, body mounted or vehicle mounted) but with any required sensing of user head/body rotation relative to the world now being done with respect to the vehicle. It is also to be noted that the audio-field rotation discussed above only concerned azimuth rotation-that is, rotation about a vertical axis. It is, of course, also possible to treat rotation of the field in elevation in a similar manner both to track head movements (nodding up and down) to achieve a selected stabilisation and to enable the user to command audio-field elevation-angle changes; appropriate modifications to the FIG. 1 apparatus to handle rotation in elevation in this way will be apparent to persons skilled in the art. Considering FIG. 1 in more detail, services are selected by subsystem 13, these services being either local (for example, an application running on a local processor) or accessible via a communications link 20 (such as a radio link or fixed wire connection providing internet or intranet access). The services can conveniently be categorised into general services such as e-mail, and services that have relevance to the immediate vicinity (augmentation services). The services are selected by selection control block 17 according to predetermined user-specified criteria and possibly also by real-time user input provided via any suitable means such as a keypad, voice input unit or interactive display. A memory 14 is used to store data about the selected services with each such service being given a respective service ID. For each selected service, memory 14 holds access data (e.g. address of service executable or starting URL) and data on the or each sound source specified by the service or user to be used to represent the service with each such sound source being distinguished by a suitable suffix to the service ID. For each sound source, the memory holds data on the or each associated audio label, each label being identified by a further suffix to the suffixed service ID used to identify the sound source. The audio labels for the selected services are either provided by the services themselves to the subsystem 13 or are specified by the user for particular identified services. The labels are preferably provided and stored in text-form for conversion to audio by a text-to-speech converter (not shown) as and when required by the spatialisation processor. Where the audio label associated with a service is to be a low-level live feed, memory 14 holds an indicator indicating this. Provision may also be made for temporarily replacing the normal audio label of a service sound source with a notification of a significant service-related event (for example, where the service is an e-mail service, notification of receipt of a message may temporarily substitute for the normal audio label of the service). As regards the full service feed of any particular service, this is not output from subsystem 13 until that service is chosen by the user by input to output selection block 12. Rather than the services to be represented in the audio interface being selected by block 17 from those currently found to be available, a set of services to be presented can be pre-specified and the related sound-source data (including audio labels) for these services stored in memory 14 along with service identification and access data. In this case, when the apparatus is in its "desktop" mode, the services in the pre-specified set of services are represented in the output audio field by the stored audio labels without any need to first contact the services concerned; upon a user selecting a service and the apparatus changing to its service mode, the service access data for the selected service is used to contact that service for a full service feed. With respect to the positioning of the service-representing sound sources in the audio field when the apparatus is in its desktop mode, each service may provide position information either indicating a suggested spatialised position in the audio field for the sound source(s) through which the service is to be represented, or giving a real-world location associated with the service (this may well be the case in respect of an augmented reality service associated with a location in the vicinity of the user). Where a set of services is pre-specified, then this position information can be stored in memory 14 along with the audio labels for the services concerned. For each service-representing sound source, it is necessary to determine its final rendering position in the output audio field taking account of a number of factors. This is done by injecting a sound-source data item into a processing path involving elements 21 to 30. This sound-source data item comprises a sound source ID (such as the related suffixed service ID) for the sound source concerned, any service-supplied position information for the sound source, and possibly also the service type (general service/augmentation service). The subsystem 13 passes each sound-source data item to a source-position set/modify block 23 where the position of the sound source is decided relative to the audio-field reference vector, either automatically on the basis of the supplied type and/or position information, or from user input 24 provided through any suitable input device including a keypad, keyboard, voice recognition unit, or interactive display. These positions are constrained to conform to the desired form (spherical or part spherical; 1D, 2D, or 3D) of the audio field. The decided position for each source is then temporarily stored in memory 25 against the source ID. Provision of a user input device for modifying the position of each sound source relative to the audio field reference, enables the user to modify the layout of the service-representing sound sources (that is, the dispositions of these sound sources relative to each other) as desired. With respect to a service having an associated real-world location (typically, an augmented reality service), whilst it is possible to position the corresponding sound source in the audio field independently of the relationship between the associated real-world location of the service and the location of the user, it will often be desired to place the sound source in the field at a position determined by the associated real-world location and, in particular, in a position such that it lies in the same direction relative to the user as the associated real-world location. In this latter case, the audio field will generally be world-stabilised to maintain the directional validity of the sound source in the audio field presented to the user; for the same reason, user-commanded rotation of the audio field should be avoided or inhibited. Positioning a sound source according to an associated real-world location is achieved in the present apparatus by a real-world location processing functional block 21 that forms part of the source-position set/modify block 23. The real-world location processing functional block 21 is arranged to receive and store real-world locations passed to it from subsystem 13, these locations being stored against the corresponding source IDs. Block 21 is also supplied on input 22 with the current location of the user determined by any suitable means such as a GPS system carried by the user, or nearby location beacons (such as may be provided at point-of-sale locations). The block 21 first determines whether the real-world location associated with a service is close enough to the user to qualify the corresponding sound source for inclusion in the audio field; if this test is passed, the azimuth and elevation coordinates of the sound source are set to place the sound source in the audio field in a direction as perceived by the user corresponding to the direction of the real world location from the user. This requires knowledge of the real-world direction of pointing of the un-rotated audio-field reference vector 42 (which, as noted above, is also the direction of pointing of the presentation reference vector). This can be derived for example, by providing a small electronic compass on a structure carrying the audio output devices 11, since this enables the real-world direction of pointing of presentation reference vector 44 to be measured; by noting the rotation angle of the audio-field reference vector 42 at the moment the real-world direction of pointing of vector 44 is measured, it is then possible to derive the real-world direction of pointing of the audio-field reference vector 42 (assuming that the audio field is being world-stabilised). It maybe noted that not only will there normally be a structure carrying the audio output devices 11 when these are constituted by headphones, but this is also the case in any mobile situation (for example, in a vehicle) where loudspeakers are involved. If the audio field is a 3D field, then as well as setting the azimuth and elevation coordinates of the sound source to position it in the same direction as the associated real-world location, block 21 also sets a range coordinate value to represent the real world distance between the user and the real-world location associated with the sound source. Of course, as the user moves in space, the block 21 must reprocess its stored real-world location information to update the position of the corresponding sound sources in the audio field. Similarly, if updated real-world location information is received from a service, then the positioning of the sound source in the audio field must also be updated. Returning to a general consideration of the FIG. 1 apparatus, an audio-field orientation modify block 26 is used to specify any required changes in orientation (angular offset) of the audio-field reference vector relative to presentation reference vector. In the present example where the audio field is to be body-stabilized and the output audio devices are headphones, the apparatus includes the afore-mentioned head tracker sensor 33 and this sensor is arranged to provide a measure of the turning of a user's head relative to their body to a first input 27 of the block 26. This measure is combined with any user-commanded field rotation supplied to a second input of block 26 in order to derive a field orientation angle that is stored in memory 29. As already noted, where headphones are used and the audio field is to be world stabilised (for example, where augmented-reality service sound sources are to be maintained in positions in the field consistent with their real world positions relative to the user), then the head-tracker sensor needs to detect any change in orientation of the user's head relative to the real world so that the audio field can be given a counter rotation. Where the user is travelling in a vehicle and the audio field is to be vehicle-stabilised, the rotation of the user's head is measured relative to the vehicle (the user's "local" world, as already noted). Each source position stored in memory 25 is combined by combiner 30 with the field orientation (rotation) angle stored in memory 29 to derive a rendering position for the sound source, this rendering position being stored, along with the source ID, in memory 15. The combiner operates continuously and cyclically to refresh the rendering positions in memory 15. Output selection block 12 sets the current apparatus mode according to user input, the available modes being a desktop mode and a service mode as already discussed above. When the desktop mode is set, the spatialisation processor 10 accesses the rendering position memory 15 and the memory 14 holding the service audio labels to generate an audio field, via audio output devices 11, in which the (or the currently-specified) audio label associated with each sound source is spatialized to a position set by the corresponding rendering position in memory 15. In generating the audio-label field, the processor 10 can function asynchronously with respect to the combiner 30 due to the provision of memory 15. The spatialisation processor 10 operates according to any appropriate sound spatialisation method, including those mentioned in the introduction to the present specification. The spatialisation processor 10 and audio output devices together form a rendering subsystem serving to render each sound source at its derived final rendering position. When the service mode is set, the full service audio feed for the chosen service is rendered by the spatialisation processor 10 according to whatever position information is provided by the service. It will be appreciated that, although not depicted, this service position information can be combined with the field orientation angle information stored in memory 29 to achieve the same stabilization as for the audio-field containing the service audio labels; however, this is not essential and, indeed, the inherent stabilization of the audio output devices (head-stabilised in the case of headphones) may be more appropriate for the full service mode. As an alternative to the full service feed being spatialised by the spatialisation processor 10, the full service feed may be provided as pre-spatialized audio signals and fed directly to the audio output devices. With the FIG. 1 apparatus set to provide a body-stabilised audio field through headphones, the user can explore the audio field in two ways, namely by turning their head and by rotating the audio field. FIG. 4 illustrates a user turning their head to explore a 2D audio field restricted to occupy part only of a spherical surface. In this case, six spatialised sound sources 40 are depicted. Of these sources, one source 40A is positioned in the audio field at an azimuth angle of X1° and elevation angle Y1° relative to the audio-field reference vector 42. The user has not commanded any explicit rotation of the audio field. However, the user has turned their head through an angle X2° towards the source 40A. In order to maintain body-stabilisation of the audio field, the audio-field reference vector 42 has been automatically rotated an angle (-;X2°) relative to the presentation reference vector 44 to bring the vector 42 back in line with the user's body straight ahead direction; the rendering position of the source relative to the presentation reference vector is therefore:
FIG. 5 illustrates, for the same audio field as represented in FIG. 4, how the user can bring the sound source 40A to a position directly ahead of the user by commanding a rotation of (-;X1°) of the audio field by user input 28 to block 26 (effected, for example, by a rotary input device). The azimuth rendering position of the sound source 40A becomes (X1°-;X1°), that is, 0°-the source 40A is therefore rendered in line with the presentation reference vector 44. Of course, if the user turns their head, the source 40A will cease to be directly in front of the user until the user faces ahead again. Audio Field Organisation-Cylindrical Field Example The FIG. 1 apparatus can be adapted to spatialize the sound sources 40 in an audio field conforming to the surface of a vertically-orientated cylinder (or part thereof). FIG. 6 depicts a general case where the audio field conforms to a notional cylindrical surface 50. This cylindrical audio field, like the spherical audio field previously described with reference to FIG. 2, is two dimensional inasmuch as the position of a sound source 40 in the field it can be specified by two coordinates, namely an azimuth angle X° and an elevation (height) distance Y, both measured relative to an horizontal audio-field reference vector 52. It will be appreciated that a 3D audio field can be specified by adding a range coordinate Z, this being the distance from the axis of the cylindrical audio field. As with the spherical audio field described above, the cylindrical audio field may be rotated (angularly offset by angle R°) relative to a presentation reference vector 54, this being done either in response to a direct user command or to achieve a particular field stabilisation in the same manner as already described above for the spherical audio field. In addition, the audio field can be axially displaced to change the height (axial offset) of the audio-field reference vector 52 relative to the presentation reference vector 54. Since it is possible to accommodate any desired number of sound sources in the audio field without over crowding simply by extending the elevation axis, there is a real risk of a "Tower of Babel" being created if all sound sources are active together. Accordingly, the general model of FIG. 6 employs a concept of a focus zone 55 which is a zone of the cylindrical audio field bounded by upper and lower elevation values determined by a currently commanded height H so as to keep the focus zone fixed relative to the assumed user position (the origin of the presentation reference vector); within the focus zone, the sound sources 40 are active, whilst outside the zone the sources 40 are muted (depicted by dashing of the hexagon outline of these sources in FIG. 6) except for a limited audio leakage 56. In FIG. 6, the focus zone (which is hatched) extends by an amount C above and below the commanded height H (and thus has upper and lower elevation values of (H+C) and (H-;C) respectively. In the illustrated example, H=0 and C is a constant; C need not be constant and it would be possible, for example, to make its value dependent on the value of the commanded height H. The general form of cylindrical audio field shown in FIG. 6 can be implemented in a variety of ways with respect to how leakage into the focus zone is effected and how a user moves up and down the cylindrical field (that is, changes the commanded height and thus the current focus zone). FIGS. 7 and 8 illustrate two possible implementations in the case where the audio field is of semi-cylindrical form (azimuth range from +90° to -;90°). In FIG. 7, leakage takes the form of the low-volume presence of sound sources 40W in upper and lower "whisper" zones 56, 57 positioned adjacent the focus zone 55. Also, the commanded height value is continuously variable (as opposed to being variable in steps). The result is that the user can effectively slide up and down the cylinder and hear both the sound sources 40 in the focus zone and, at a lower volume, sound sources 40W in the whisper zones. In FIG. 8, the service sound sources are organised to lie at a number of discrete heights, in this case, four possible heights effectively corresponding to four "floors" here labelled "1" to "4". Preferably, each "floor" contains sound sources associated with services all of the same type with different floors being associated with different service types. The user can only command step changes in height corresponding to moving from floor to floor (the extent of the focus zone encompassing one floor). Leakage takes the form of an upper and lower advisory sound source 60, 61 respectively positioned just above and just below the focus zone at an azimuth angle of 0°. Each of these advisory sound sources 60, 61 provides a summary of the services (for example, in terms of service types) available respectively above and below the current focus zone. This permits a user to determine whether they need to go up or down to find a desired service. It will be appreciated that the forms of leakage used in FIGS. 7 and 8 can be interchanged or combined and that the FIG. 8 embodiment can provide for sound sources 40 on the same floor to reside at different heights on that floor. It is also possible to provide each floor of the FIG. 8 embodiment with a characteristic audio theme which rather than being associated with a particular source (which is, of course, possible) is arranged to surround the user with no directionality; by way of example, a floor containing museum services could have a classical music theme. In arranging for the FIG. 1 apparatus to implement a cylindrical audio field such as depicted in any of FIGS. 4-6, the positions set for the sound sources by block 23 are specified in terms of the described cylindrical coordinate system and are chosen to conform to a cylindrical or part-cylindrical organisation in 1, 2, or 3D as required. The orientation and vertical positioning of the audio field reference vector 42 are set by block 26, also in terms of the cylindrical coordinate system. Similarly, combiner 30 is arranged to generate the sound-source rendering positions in terms of cylindrical coordinates. The spatialisation processor must therefore either be arranged to understand this coordinate system or the rendering positions must be converted to a coordinate system understood by the spatialisation processor 10 before they are passed to the processor. This latter approach is preferred and thus, in the present case, assuming that the spatialisation processor is arranged to operate in terms of the spherical coordinate system illustrated in FIG. 2, a converter 66 (see FIG. 9) is provided upstream of memory 15 to convert the rendering positions from cylindrical coordinates to spherical coordinates. Whilst it would be possible to use a single coordinate system throughout the apparatus regardless of the form of audio field to be produced (for example, the positions of the sound sources in the cylindrical audio field could be specified in spherical coordinates), this complicates the processing because with an appropriately chosen coordinate system most operations are simple additions or subtractions applied independently to the individual coordinates values of the sound sources; in contrast, if, for example, a spherical coordinate system is used to specify the positions in a cylindrical field, then commanded changes in the field height (discussed further below) can no longer simply be added/subtracted to the sound source positions to derive their rendering heights but instead involve more complex processing affecting both elevation angle and range. Indeed, by appropriate choice of coordinate system for different forms of audio field, equivalent operations with respect to the fields translate to the same operations (generally add/subtract) on the coordinate values being used so that the operation of the elements 25, 26, 29 and 30 of the apparatus is unchanged. In this case, adapting the apparatus to a change in audio-field form, simple requires the block 23 to use an appropriate coordinate system and for converter 66 to be set to convert from that coordinate system to that used by the spatialisation processor 10. With respect to adaptation of the FIG. 1 apparatus to provide the required capability of commanding changes in height for the cylindrical audio field systems illustrated in FIGS. 4-6, such height changes correspond to the commanding of changes in the elevation angle already described for the case of a spherical audio field. Thus, a height change command is supplied to the block 26 to set a field height value (an axial offset between the field reference vector and the presentation reference vector) which is then combined with the elevation distance value Y of each sound source to derive the elevation value for the rendering position of the source. As regards how the focus zone and leakage features are implemented, FIG. 9 depicts a suitable variation of the FIG. 1 apparatus for providing these features. In particular, a source parameter set/modify block 70 is interposed between the output of combiner 30 and the converter 66. This block 70 comprises one or more units for setting and/or modifying one or more parameters associated with each sound source to condition how the sound source is to be presented in the audio field. As will be seen hereinafter with respect to the FIG. 10 apparatus, the block 70 can include a range of different type of units that may modify the rendering position of a source and/or set various sounding effect parameters for the source. In the present case, the block 70 comprises a cylindrical filter 71 that sets a audibility (volume level) sounding-effect parameter for each sound source. The set parameter value is passed to memory 15 for storage along with the source ID and rendering position. When the spatialisation processor comes to render the sound source audio label according to the position and audibility parameter value stored in memory 15, it passes the audibility value to a sounding effector 74 that conditions the audio label appropriately (in this case, sets its volume level). In the case of the FIG. 7 arrangement, the cylinder filter 71 is responsive to the current field height value (as supplied from memory 29 to a reference input 72 of block 70) to set the audibility parameter value of each sound source: to 100% (no volume level reduction) for sound sources in the focus zone 55; to 50% for sound sources in the "whisper" zones 56 and 57; and to 0% (zero volume) for all other sound sources. As a result, the sounding effector 74 mutes out all sound sources not in the focus or whisper zones, and reduces the volume level of sound sources in the whisper zones. In the case of the FIG. 8 arrangement, the cylinder filter 71 performs a similar function except that now there are no whisper zones. As regards the upper and lower advisory sound sources 60 and 61, the subsystem 13 effectively creates these sources by:
The source IDs passed to the block 23 are there associated with null position data before being passed on via memory 25 and combiner 30 to arrive at the cylinder filter 71 of block 70. The filter 71 recognises the source IDs as upper and lower advisory sound source IDs and appropriately sets position data for them as well as setting the audibility parameter to 100% and setting a parameter specifying which summary audio label is appropriate for the current floor. This enables the spatialisation processor to retrieve the appropriate audio label when it comes to render the upper or lower advisory sound source. It will be appreciated that partially or fully muting sound sources outside of a focus zone can also be done where the apparatus is set to generate a spherical audio field. In this case, the apparatus includes blocks 70 and 74 but now the cylinder filter 71 is replaced by a "spherical filter" muting out all sound sources beyond a specified angular distance from a current facing direction of the user. The current facing direction relative to the presentation reference vector is derived by block 26 and supplied to the filter 71. It may be noted that in the case where the audio output devices 11 are constituted by headphones, the direction of facing of the user corresponds to the presentation reference vector so it is a simple matter to determine which sound sources have rendering positions that are more than a given angular displacement from the facing direction. Along with the implementation of a focus zone for a spherical audio field, it is, of course, also possible to provide the described implementations of a leakage feature. Multiple Audio Sub-fields FIG. 10 shows a second apparatus for producing an audio field to serve as an audio interface to services. This apparatus is similar to the FIG. 9 variant of the first apparatus but provides for multiple audio "sub-fields" and has a variety of sound-source parameter conditioning units for facilitating a clear audio presentation. Elements of the first and second apparatus that have similar functionality have been given the same reference numerals and their description will not be repeated below for the second apparatus except where there is modification of functionality to accommodate features of the second apparatus. The second apparatus, like the first apparatus, is capable of producing (part) spherical or part (cylindrical) 1D, 2D or 3D audio fields (or, indeed, any other form of audio field) according to the positions set for the sound sources by block 23. As mentioned, the FIG. 10 apparatus provides for multiple "sub-fields". Each sub-field may be considered as an independent audio field that can be rotated (and, in the case of a cylindrical field, vertically re-positioned) by changing the offset between the presentation reference vector and an audio-field reference specific to the sub-field. Further, each sub-field can have a different stabilization set for it-thus, for example, sound sources representing general services can be assigned to a head-stabilised sub-field whilst sound sources representing augmented-reality services can be assigned to a world-stabilised sub-field. The rotation/displacement of each sub-field and the setting of its stabilization is done by block 26 with the resultant values being stored in memory 29. Whether or not the block 26 modifies the azimuth-angle value of a sub-field to reflect a sensed rotation of the user's head will thus depend on the stabilization set for the sub-field and, as already described, on whether the audio output devices are head-mounted, body-mounted, vehicle-mounted or fixed with respect to the world (or, in other words, whether the presentation reference vector is head, body, vehicle or world stabilised). To add flexibility to the FIG. 10 apparatus, the current stabilisation of the presentation reference vector is fed to the block (see arrow) to enable the latter to make any appropriate changes to the sub-field orientations as the user turns (and/or nods) their head. Each service sound source is assigned by block 23 to a particular sub-field and an identifier of its assigned sub-field is stored with the source ID in memory 25 along with the position of the sound source relative to the audio-field reference associated with the assigned sub-field. The combiner 30 is supplied from memory 29 with the rotation/displacement values of each sub-field and for each service sound source combines the values of the related sub-field with the sound-source coordinate values; as a result, each sound source is imparted the rotations/displacements experienced by its sub-field. For each service sound source, the output of the combiner comprises source ID, position data, and sub-field identifier. As will be seen below, assigning sound sources to different sub-fields may be done for reasons other than giving them different stabilizations; for example, it may be done to identify a group of service sound sources that are to be subject to a particular source-parameter modification process in block 70. It should also be noted that different sub-fields may have different dimensions and even different forms so that one sub-field could be a 2D spherical surface whilst another sub-field could be of 3D cylindrical form. Facilitating Clear Presentation As well as the cylindrical filter 71, the source parameter set/modify block 70 includes a number of sound-source parameter conditioning units 80 to 85 for facilitating a clear audio presentation. The function of each of these units will be described more fully below. It is to be understood that the units need not all be present or operational together and various combinations of one or more units being concurrently active are possible; however, not all combinations are appropriate but this is a matter easily judged and will not be exhaustively detailed below. Also, certain units may need to effect their processing before others (for example, units that affect the final rendering position of a sound source need to effect their processing before units that set sounding effect parameters in dependence on the final rendering position of a sound source); again, it will generally be apparent when such ordering issues are present and what ordering of the units is required to resolve such issues and an exhaustive treatment of these matters will not be given below. Unit 80 is a focus expander that serves to modify the rendering positions of the sound sources to spread out the sound sources (that is, expand or dilate the audio field) in azimuth in the region of the current direction of facing of the user (or other appropriate direction) in order to facilitate discrimination between sound sources. Referring to FIG. 11, this shows a field of 180° extent in azimuth with the user currently facing in the direction of the audio-field reference vector 90. The focus expander 80 operates to linearly expand the 15° segments 92 on both sides of the facing direction 91 into respective 45° segments 93 (see the hatched zones). The remaining segments are correspondingly compressed to maintain an overall 180° azimuth range-in this case, this results in two 75° segments 94 being compressed into respective 45° segments 95; as an alternative (not illustrated), the remaining segments could simply be angularly displaced from their normal positions without compressing them. For sub-fields that are head-stabilised, turning of the user's head does not change the 15° segments subject to expansion; however, azimuth rotation of such a sub-field does result in the expansion being applied to different segments of the sub-field. For sub-fields that are not head-stabilised, as the user turns their head, the segments subject to expansion change. This is illustrated in FIG. 12 where a user has turned to the right 75° relative to the audio-field reference vector of a body-stabilised audio sub-field with an initial ±90° range either side of the reference vector. This results in the most clockwise 30° of the original field (segments 92) being expanded (symmetrically with respect to the facing direction) so that now the audio sub-field extends round further in the clockwise direction than before. The remaining 150° segment 97 of the original audio sub-field is expanded into a 90° segment 98. In order for the focus expander 80 to effect the required processing of the azimuth rendering positions of the sound sources, it is supplied (input 78 to block 70) with the angle of the facing direction relative to the current presentation reference vector, this angle being determined by the block 26 in dependence on the current stabilization of the presentation reference vector and the sensed head rotation. Of course, where the presentation reference vector is head-stabilized (i.e. headphones are being used), the angle between the facing direction and the presentation reference vector will be zero; in other cases it will generally correspond to the angle measured by the head-tracker sensor 33. Given the facing direction angle relative to the presentation reference vector, and bearing in mind that the sound-source positions supplied to block 70 are relative to that vector, it is a straightforward matter for the focus expander 80 to determine which sound sources lie within the segments 92 and then make the required changes to the azimuth values of the sound-source rendering positions of these sources in order to achieve the desired audio-field dilation; similarly, the rendering positions of the other sound sources are adjusted as required. It will be appreciated that the user can be enabled to turn the focus expander 80 on and off as desired. It is also possible to arrange for the focus expander to be applied only to one or more selected sub-fields rather than to all fields indiscriminately. Furthermore, whilst the focus expander has been described above as operating on azimuth angles, it could additionally or alternatively be caused to act on the elevation coordinate values (whether angles or distances). Again, whilst the expansion has been described above as being uniform (linear), it could be applied in a non-linear manner such that a larger expansion is applied adjacent the facing direction than further away. The angle of application of the expansion effect can also be made adjustable. Rather than the focus expander 80 expanding a region of the audio field set relative to the current facing direction, the focus expander can be arranged to expand a region set relative to some other direction (the 'focus reference direction'), such as a specific world-stabilised direction or the presentation reference vector. In this case, the focus expander is provided with appropriate information from block 26 to enable it to determine the relative offset between the focus reference direction and the presentation reference vector (this offset being, of course, zero if the focus reference direction is set to be the presentation reference vector). Arrow 79 in FIG. 10 generally represents user input to block 70 whether for controlling the focus expander 80 or any other of the units of the block. How the user input is derived is an implementation detail and may, for example, be done by selection buttons, a graphical user interface, or voice command input subsystem. Unit 81 of the source-parameter set/modify block 70 is a segment muting filter 81 that is operative to change the audibility state of sound sources in user-specified segments of one, some or all the audio sub-fields (a default of all sub-fields is preferably set in the filter 81 with the possibility of the user changing this default). In particular, the segment muting filter changes the audibility state of segment sound sources (in either direction) between un-muted and at least partially muted by appropriately setting the value of an audibility (sound volume) parameter of the sound sources. FIG. 13 illustrates the effect of the segment muting filter in respect of an audio sub-field, of 180° azimuth extent, shown developed into a rectangular form 100 and with spatialised sound sources 40. In this example, the audio field is divided into five segments relative to the audio-field reference vector, namely:
The filter 81 acts to change the audibility parameter of each sound source in a segment back and forth between 100% and 0% (or a preset low level) in response to user input. Preferably, speech form input is possible so that to mute sound sources in segment 102, the user need only say "Mute Left" (FIG. 13 depicts these sounds sources as muted by showing them in dashed outline). To bring back these sound sources to full volume, the user says "Un-Mute Left". As already described with respect to the cylindrical filter 71, the sound volume specified by the audibility parameter is implemented by sounding effector 74, the effector being passed the parameter when the spatialisation processor 10 requests to be supplied with the sound label for the sound source concerned. Preferably, the segments can be muted and un-muted independently of each other. An alternative is to arrange for only one segment to be muted at a time with the selection for muting of a segment automatically un-muting any previously muted segment; the opposite is also possible with only one segment being un-muted at a time, the un-muting of a segment causing any previously un-muted segment to be muted. It is also possible to arrange for several segments to be muted simultaneously in response to a single command-for example, both the "left" and "far left" segments 102, 103 in FIG. 13 could be arranged to be muted in response to a user command of "Mute All Left". The segments are pre-specified in terms of their azimuth angular extent relative to the audio-field reference vectors by segmentation data stored in the segment muting filter or elsewhere. In order for the segment muting filter to mute the sound sources corresponding to a segment to be muted, the filter needs to know the current azimuth angle between the audio field reference vectors and the presentation reference vector since the sound-source azimuth angles provided to the filter are relative to the latter vector. The required angles between the audio-field and presentation reference vectors is supplied on input 76 from block 26 to block 70. As an alternative to the segments being specified relative to the audio-field reference vectors, the segments can be specified relative to the facing direction of the user (which may, in fact, be more natural). In this case, the segment muting filter needs to know the angle between the current facing direction and the presentation reference vector; as already described, this angle is provided on input 78 to block 70. A further alternative is to pre-specify the segments relative to the presentation reference vector (which, of course, for headphones is the same as specifying the segments relative to the user's facing direction). Whilst segment muting has been described using segmentation in azimuth, it will be appreciated that the segmentation can be effected in any appropriate manner (for example, in azimuth and elevation in combination) and the term 'segment' is herein used without any connotation regarding the form or shape encompassed. Rather than a segment remaining muted until commanded to return to its un-muted state, a muted segment can be arranged only to stay muted for a limited period and then to automatically revert to being un-muted. Unit 82 is a cyclic muting filter. As depicted in FIG. 14 (which uses the same field development as FIG. 13), this filter 82 works on the basis that the sound sources 40 are divided into groups 110 to 114 and the filter 82 operates cyclically to change the audibility state of the sound sources so as to at least partially mute out all but one group of sources in turn-in FIG. 14, all groups except group 111 are currently muted. The un-muted group remains un-muted, for example, for 10 seconds before being muted (partially or fully) again. As with the segment muting filter, the filter 82 operates by setting the value of an audibility parameter of each sound source. Rather than requiring a group ID to be assigned to each sound source and transferred along with the sound-source ID, position data, and sub-field identifier to the block 70, grouping can be achieved by assigning a separate sub-field for each group. The grouping of sound sources can be effected automatically by service type (or more generally, one or more characteristics associated with the item represented by the sound source concerned). Alternatively, the grouping of the sound sources can be effected automatically according to their positions in the audio field (possibly taking account their relation to the presentation reference vector, the audio field reference vectors, or user direction of facing). A further possibility is for the grouping to be user specified (via block 23). In one possible grouping arrangement, each sound source is assigned to a respective group resulting in each sound source being un-muted in turn. Preferably, the user can also specify that one or more groups are not subject to cyclic muting. Additionally, the user can be given the option of setting the un-muted duration for each group. As already indicated, muted groups need not be fully muted. Where the sound sources are assigned to groups according to their positions, a possible muting pattern would be to fully mute sound sources in groups lying either side of the currently un-muted group of sources, and to partially mute the sound sources of all other groups. Rather than the un-muting and muting of the groups being effected in an abrupt manner, the group whose limited period of being un-muted is ending can be cross-faded with the group whose period of being un-muted is next to occur. Unit 82 is a collection collapser the basic purpose of which is to respond to a predetermined user command to collapse all sound sources that are members of a specified collection of sound sources to a single collection-representing sound source at a particular location (which can be head, body, vehicle or world stabilised). The member sound sources of the collection can be identified by a specific tag associated with each sound source ID; however, it is convenient to assign all sound sources to be collapsed to the same sub-field and simply rely on the sub-field ID to identify these sources to the block 70. FIG. 15 illustrates the general effect of the collection collapser 82 for a situation where all augmented-reality sound sources 40[AR] are members of the same collection and have been assigned to the same world-stabilised sub-field; these augmented-reality sound sources are arranged to be collapsed to a single collection-representing sound source 120 positioned at the top center of the audio sub-field. Other positions for the source 120 are, of course, possible such as in line with the current direction of facing or the location of a particular one of the sound sources being collapsed. The collection collapser is further arranged to reverse the collapsing upon receipt of a suitable user command. The collection-representing sound source 120 will generally not be present when the member sound sources of the collection are un-collapsed though it is possible to leave the collection-representing sound source un-muted to serve, for example, as notification channel to inform the user of events relevant to the collection as a whole. In a typical implementation, the collection-representing sound source is created by the subsystem 13 and is given an ID that indicates its special role; this sound source is then assigned to the same sub-field as the collection member sound sources to be collapsed. The collection-representing sound source is also given its own audio label stored in memory 14 with this label being arranged to be temporarily substituted for by any notifications generated in relation to the collection member sound sources (each sound source is also arranged to have its normal label temporarily replaced by any notification related to that source). Whilst the collection member sound sources are not collapsed, the audibility parameters of these sound sources remain at 100% but the collection-representing sound source has its audibility parameter set to 0% by the collection collapser. However, when the collection collapser 83 is triggered to collapse the collection member sound sources, these sources have their audibility parameters set to 0% whilst that of the collection-representing source is set to 100% thereby replacing the collapsed sources with a single sound source emitting the corresponding audio label (potentially periodically interrupted by notifications from the services associated with the collapsed sources). On user command, the collapsed sound sources are un-muted and the collection-representing sound source muted, thereby restoring the collection to its un-collapsed state. Rather than the collection changing from its un-collapsed state to its collapsed state in response to user command, the collection collapser can be arranged to effect this change automatically-for example, if there has been no activity in respect of any member sound source (user service request/service-originating event notification) for a predetermined period of time, then the collection collapser can be arranged to automatically put the collection in its collapsed state. Similarly, the collection collapser can automatically un-collapse the collection in response, for example, to the receipt of more than a threshold number of service event notifications within a given time, or upon the user entering a particular environment (in the case of a mobile user provided with means for detecting the user's environment either by location or in some other manner). To provide clear feedback to the user as to what is occurring when the collection is being collapsed and un-collapsed, the collection collapser is preferably arranged to change the collection between its two states non-instantaneously and with the accompaniment of appropriate audible effects. For example, during collapse, the collection-representing sound source can be faded up as the collection-member sound sources are faded out. This can be accompanied by a sound such as a sucking in sound to indicate that the member sound sources are notionally being absorbed into the collection-representing sound source. Alternatively, the locations of the member sound sources can be moved over a second or two to the location of the collection-representing sound source. The reverse effects can be implemented when the collection is un-collapsed. It may in certain circumstances to have more than one collection-representing sound source associated with a collection. As regards the non-collection sound sources (if any) in the audio field, these are typically left un-disturbed by changes in the state of the collection. However, it would alternatively be possible to arrange for such sound sources to be modified to adapt to the presence or absence of the collection member sound sources. For example, upon un-collapsing of the collection, the location of any sound source close to where a member sound source appears in the audio field can be changed to ensure a minimum separation of sound sources. As another example, upon un-collapsing of the collection the other sound sources can be partially muted, at least temporarily. It will be appreciated that the collection collapser provides more than just a way of opening an audio menu where the member sound sources represent menu list items; in particular, the distribution of the collection member sound sources in the un-collapsed collection is not constrained to that of a list but is determined by other considerations (for example, where the sound sources represent augmented reality services, by the real-world locations of these services). Unit 84 is a sub-field sound setter intended to set a sounding effect parameter in respect of sound sources of a particular sub-field or sub-fields. The sound setter is operative to set a particular sounding effect parameter as either on or off for each sound source, whilst the sounding effector 74 is arranged to apply the corresponding sound effect to all sound sources for which the parameter is set to on. Preferably, as default, when the sound setter is enabled the sound sources of all sub-fields have the related sounding effect parameter set to on; however, the user can de-select one or more sub-fields for this treatment, as desired. In fact, multiple different sound setters 84 can be provided, each associated with a different sound effect. Typical sound effects are volume or pitch modulation, frequency shifting, distortion (such as bandwidth limiting or muffling), echo, addition of noise or other distinctive sounds, etc. One reason to employ the sound setter 84 is to make it easy to distinguish one type of service from another or to distinguish the synthesised sound sources from real sound sources in the environment. In this latter case, the audio output devices are, of course, configured to permit the user to hear both real-world sounds as well as the synthesised sounds. The user is preferably enabled to choose, via appropriate input means, what sound effect is to be used to make the synthesised sounds distinct; advantageously, the user can also choose to apply or remove the selected sound effect. In fact, another way of distinguishing between one group of sounds and another (such as real and synthesised sounds) is by way of specifying a particular stabilization for a sub-field(s) containing one of the group of sound sources to be distinguished. Thus, audio labels for augmented-reality services can be distinguished from real world sounds by assigning the audio-label sound sources to a head-stabilised field so that they move relative to the real world as the user turns their head. As another example, the audio labels of general services could be assigned to a head-stabilised sub-field and the audio labels of augmented-reality services to a world-stabilised sub-field. As a refinement to always applying the same stabilization to a particular sub-field, the block 26 can be arranged to apply a stabilization scheme in which the sub-field is only updated periodically to a specified underlying stabilization, no account being taken between updates of any changes in orientation of the user's body or head (thereby automatically applying the stabilization associated with the presentation reference vector between updates). Unit 85 is a range sound setter and is applicable only where an audio sub-field has depth (that is, the range parameter can be different for different sound sources of the sub-field). The range sound setter, when enabled in respect of a sub-field, is operative, for each sound-source in the sub-field, to set a sound source parameter according to the range of the sound source. The purpose of doing this is to impart an audible characteristic to the sound source that indicates to the user at least a general range of the sound source. This parameter could, for example, be the audibility parameter with the value of this parameter being set such that sound sources at a greater range are presented at a lower volume. However, in a preferred embodiment, the value of the parameter controlled by unit 85 is used to select which audio label to render from a set of audio labels associated with a sound source, each label having a different presentation character at least one aspect of which, other than or additional to loudness, differs between labels. This aspect is, for example, speaking style, vocabulary, speaker voice, etc. The mere change in a range value included in an announcement is not considered to be a change in the presentation character of the announcement. The user can readily learn to associate the differing presentation characters with particular range bands. FIG. 16 illustrates an example concerning a sound source for an augmented-reality notification service from the user's local newspaper shop; this service sound source has three associated audio labels, stored for it in memory 14, of increasing familiarity the closer the sound source is to the user: | ||||||