Sound Projector Set-up

  • Published: May 7, 2009
  • Family: 7
  • Cited Works: 1
  • Cited by: 6
  • Cites: 0
  • Additional Info: Full text

SOUND PROJECTOR SET-UP

The present invention concerns a device including a plurality of acoustic transducers. The device is capable of producing beams of audible sound. In particular, the invention is directed towards such devices when used for home entertainment or professional sound reproduction applications. In even more particularity, the invention concerns methods and systems useful for configuring (i.e. setting up) such devices.

WO 01/23104, WO 2/078388, WO 03/034780, WO 03/059005, WO 03/071827, WO 2004/075601, WO 2005/027514, WO 2005/086526, WO 2006/003528,

WO 2007/007083, the disclosure of each of which is hereby incorporated by reference, describe systems comprising an array of transducers and their use to achieve a variety of sound effects. These documents describe methods and apparatus for taking an input signal, replicating it a number of times and modifying (e.g. delaying) each of the replicas before routing the modified replicas to respective output transducers such that a desired sound field is created. This sound field may comprise, inter alia, a directed, steerable, beam, a focussed beam or a simulated origin. The methods and apparatus of the above and other related applications are referred to in the following as "Sound Projector" technology. The teachings of the above-mentioned documents may be used to implement aspects of the present invention.

Conventional surround-sound is generated by placing individual loudspeakers at appropriate positions surrounding the listener's position. Typically, a surround sound system employs a left, centre and right speaker located in the front half space and two rear speakers located in the rear half space. The terms "front", "left", "centre", "right" and

"rear" are used relative to the listener's position and orientation. A subwoofer is also often provided, and it is usually specified that the subwoofer can be placed anywhere in the listening environment.

These known surround sound systems receive input audio information and decode the information according to known standards (such a Dolby Surround 5.1 or Digital Theatre Sound - DTS). The decoded information provides output audio streams for the various audio channels, with each audio channel usually being emitted through one loudspeaker or a combination of two speakers. The audio information can itself comprise the information for each of the several channels (as in Dolby Surround 5.1) or for only some of the channels, with other channels being simulated (as in Dolby Pro-Logic systems).

These conventional surround sound systems suffer the problem that a large number of separate speaker units are required and these units must be distributed around the room. The speaker units need to be connected together, most commonly by wires, which can be inconvenient and unsightly for a user.

These problems were addressed in WO 01/23104 and WO 02/078388 by the provision of a Sound Projector that can generate a surround-sound environment using only a signal standalone bank of output transducers. The described system uses an array of output transducers to simultaneously direct a plurality of sound beams, with each beam representing one of the surround sound channels, with such sound beams being reflected from surfaces such as the ceiling and walls back to the listener. The listener perceives the sound beam as if it was emitted from an acoustic mirror image of a source located at or behind the spot where the last reflection took place. In this system, a true, rather than a simulated, surround sound effect is achieved because, as far as the listener is concerned, the sound really does come from the simulated origin, which can be behind, as well as to the side and in front of the listener. Accordingly, a surround sound system having the advantage of being formed of only a single unit in the room was disclosed.

The surround sound effect is dependent upon the beams reaching the listener from the correct direction. This in turn is dependent on the room geometry and so it is necessary with this system for the apparatus to be correctly set up such that the various sound beams are emitted in the correct directions to achieve the surround sound effect. Furthermore, it is important in a surround sound system that temporally corresponding parts of the different channels reach the user at the same time. As the sound beam channels follow different path lengths, it is necessary to relatively delay some channels compared to others in order to ensure proper temporal correspondence at the user's position (see WO 02/078388). Accordingly, it was necessary in the prior art for the Sound Projector systems to be programmed by experts to ensure that the geometry of the particular room was taken into account.

There remains a desire to facilitate the set-up procedure such that untrained personnel (such as the average user) can easily set up the surround sound system without the need for expert installation.

WO 2004/066673 discloses two methods for setting up a Sound Projector system. The methods disclosed rely on initially using the Sound Projector to provide a model of the room environment. Once the room environment model has been calculated, choices can be made about where to direct the sound beams such that they reach the listener from the appropriate direction and at the appropriate time.

hi the first disclosed embodiment of WO 2004/066673, the Sound Projector is provided with a microphone located at the centre of the Sound Projector array. Pulses of sound are emitted from the array in chosen directions and the microphone is used to detect the reflected, refracted and diffracted return sounds. The time is measured between emitting a pulse and receiving the first reflected pulse back. This embodiment works on the principal that all objects, whatever their orientation, will reflect some of the sound pulse precisely back in the direction from which it came. Accordingly, an indication of the distance to the nearest object in each direction can be obtained. This is used to provide a computer model of the room in which the Sound Projector is located. Various processing procedures are necessary to account for room corners, room apertures (such as doorways and windows) and pieces of furniture within the room that could disrupt the sound beam path. A lot of processing power is required and the system can still result in an inaccurate room model being provided. An inaccurate room model can mean that the sound beams selected may not be reflected as they should and the surround sound effect can in many cases be lost.

The second disclosed embodiment of WO 2004/066673 uses a microphone at the listening position and uses one of the transducers in the array to output a sound pulse. No sound beams are formed in this embodiment. A fuzzy clustering algorithm is used to estimate the room geometry. -A-

Both of the embodiments of WO 2004/066673 suffer from the problem that the set-up procedure is predicated on estimating the geometry of the room (i.e. the position of the walls, ceiling etc). Any error in determining the geometry of the room can have highly adverse consequences for the surround sound produced by the resulting system. For example, if a wall is mis-positioned by even just a few centimetres, or is positioned correctly but deemed to have an angle (with respect to the Sound Projector) that is inaccurate by a few degrees, then the sound beams can take a very different path than predicted, hi general, all of the walls and ceiling positions/orientations predicted by the method will have some error and these errors will accumulate to provide that the sound beams do not, in general, follow the intended path. Furthermore, creating and storing a computer model of the listening environment requires high processing power and a lot of memory, especially given the numerous different shapes of listening environment that are possible.

As such, it would be desirable to devise a method for setting up the Sound Projector system that does not involve forming an acoustic model of the listening environment.

Contrary to the teachings of WO 2004/066673, the present invention stems in part from the realisation that better Sound Projector set-up can be achieved without creating a model of the listening environment. Accordingly, the invention provides that an input transducer

(e.g. a microphone) at the listening position is used to "listen" to a test signal as it is swept in a beam through a range of angles. From this information the system can determine those angles where the sound beam reaches the input transducer with maximum intensity. Furthermore, the path length of the sound beams can be determined by measuring the time from emitting the test signal to receiving the maximum reflection of the test signal. Methods and apparatus are disclosed for identifying the appropriate reflection from a received signal that can be the result of a very complicated superposition of a lot of reflections arriving at different times.

As the invention works by determining the angles for which a good reflection is achieved, the resulting surround sound system provides a good surround sound effect at the listening position. Furthermore, the computational complexity of calculating and storing a model of the listening space is avoided and the errors that this method can introduce are thereby also avoided. The method and system of the present invention also allows the path length of the various beams to be determined, thereby assisting in selecting which channels are directed in which directions and also assisting in providing the data necessary for the Sound Projector to ensure that, in use, the channels reach the listening position time- synchronously. The path length information is also extremely useful (preferably in conjunction with the angle and maximum correlation data) in selecting the most reliable possible beam directions.

The method and apparatus of the present invention, by its very nature, need not concern itself with extraneous information that the skilled person would learn is important from WO 2004/066673, such as the geometry of the room, the nature of the reflective surfaces in the environment (e.g. the wall materials, ceiling materials and coverings), the locations of sound reflective and/or sound obstructive obstacles in the environment (e.g. furniture) or the locations of sound "apertures" (such as doors and windows). The present invention takes the listening environment as it is and identifies the paths that are the best. Accordingly, all of the above factors are automatically taken into account without requiring any special processing.

The set-up method of the invention can use an input transducer, such as a microphone, at the listening position which detects specific test signals emitted from the Sound Projector. The method can use the detected signals to compute the optimum beam parameters, allowing optimum set-up of the Sound Projector, for example for surround sound listening.

In a first aspect, the invention provides a method for use in setting up a speaker system that comprises an array of output transducers capable of generating plural directed beams of sound, said method comprising: emitting a test signal in the form of a directed sound beam from the speaker system into a room; detecting a reflection of said sound beam test signal at a location in said room; and determining a time of a maximum correlation between said emitted sound beam test signal and said reflected sound beam test signal.

The sound beam is preferably emitted at an angle into the room. The determined time of maximum correlation is preferably used to calculate a path length for said directed sound beam.

The time of maximum correlation is preferably obtained by convolving the reflected test signal detected at said location in the room with the emitted test signal and determining the time at which said maximum correlation occurs.

The method can advantageously be repeated for a plurality of angles and the magnitude of the maximum correlation between said emitted sound beam test signal and said reflected sound beam test signal is determined for each of said plurality of angles.

The test signal is usefully in the form of a directed sound beam that is swept through a range of angles.

The method preferably further comprises storing, for each emission angle of said directed sound beam test signal, the time of said maximum correlation and preferably also the magnitude of said maximum correlation.

Angles are generally measured with respect to a line passing through the centre of said speaker system and being perpendicular to the front face of said speaker system.

The method may further comprise initially determining the position of said location in said room relative to said speaker system by detecting direct, non-reflected, test signals at said location in said room.

A sound beam direction for a centre channel is preferably selected, which beam direction points from said speaker system directly towards said location in said room.

A sound beam direction for a centre channel is preferably selected to be the direction for which a maximum correlation exists between said emitted sound beam test signal and a directly received sound beam test signal at said location in said room. Those angles for which a peak of maximum correlation exists are preferably identified.

A peak is typically defined as existing for an angle at which the maximum correlation is greater than the maximum correlation for either of the two neighbouring angles.

The determined time of maximum correlation can be used to select the angle of at least one of the surround sound channels.

A sound beam direction for each of a left sound channel and a right sound channel is typically selected, which direction corresponds to an angle at which there is a peak magnitude of maximum correlation to the left and right of the centre channel, respectively.

A sound beam direction for a left sound channel is usefully selected, which direction corresponds to the angle at which there is the largest peak magnitude of maximum correlation out of all of the interrogated angles to the left of the centre channel.

A sound beam direction for a right sound channel is usefully selected, which direction corresponds to the angle at which there is the largest peak magnitude of maximum correlation out of all of the interrogated angles to the right of the centre channel.

The method preferably determines if the peak magnitude of maximum correlation for the left sound channel is smaller than a predetermined percentage of the peak magnitude of maximum correlation for the centre channel, and then switches the speaker system to "direct mode".

The method preferably determines if the peak magnitude of maximum correlation for the right sound channel is smaller than a predetermined percentage of the peak magnitude of maximum correlation for the centre channel, and then switches the speaker system to "direct mode".

The selected sound beam direction for the left surround sound channel preferably correspαώδ>to an angle at which there is a peak magnitude of maximum correlation to the left of the centre channel and which direction has a smaller beam angle and larger time of maximum correlation than the selected direction for the left sound channel.

The selected sound beam direction for the right surround sound channel preferably corresponds to an angle at which there is a peak magnitude of maximum correlation to the right of the centre channel and which direction has a smaller beam angle and larger time of maximum correlation than the selected direction for the right sound channel.

The method preferably further comprises determining a plurality of candidate beam directions for the surround sound channels and assigning a quality coefficient to each candidate.

The quality coefficient is initially preferably determined by multiplying the maximum correlation value for the beam angle by the beam path length although the path length used when calculating the quality coefficient is preferably limited to some multiple of the path length calculated for the left sound channel or right sound channel, respectively.

The method may include disregarding candidate beams. Such candidate beams can be disregarded, preferably by setting the quality coefficient to be zero, if any of the following conditions (a) to (e) exist:

(a) the magnitude of maximum correlation for the candidate beam direction is less than a predetermined percentage of the magnitude of maximum correlation for the beam direction of the centre channel;

(b) the candidate beam has a larger angle and a larger time of maximum correlation than an already selected left or right sound channel;

(c) the candidate beam has a smaller angle and a smaller time of maximum correlation than an already selected left or right sound channel;

(d) the candidate beam has a time of maximum correlation within a predetermined amount (e.g. 100 - 500 μs, preferably 200 μs) and an emission angle within a predetermined amount (e.g. 5 - 25°, preferably 15°) of an already determined candidate beam;

(e) the candidate beam is directed in a direction where the path length for adjacent angles differs by a predetermined amount from the candidate beam path length.

Usually, the candidate beam direction having the largest quality coefficient is selected for a surround sound channel.

The focal length for the centre channel can be advantageously set to be a fixed negative number, such as -0.5 m.

The focal length for the left sound channel and/or the right sound channel can preferably be set based on the path lengths for those channels, preferably being set at one half of the path length.

A correlation can be calculated only for those sound beams having a path length larger than the distance from the speaker system to the location in said room.

If the angle between the identified left sound channel and identified right sound channel is smaller than a predetermined angle, a "corner mode" can be set for the speaker system. When a "corner mode" is set, the left sound channel and right sound channel can be emitted in the "direct mode", hi the "direct mode", the left sound channel and right sound channel can be directed in the same direction as the centre sound channel and a subset of the transducers in the array can be used to emit the left sound channel and right sound channel. Preferably, in addition, in direct mode, the focal length of the left sound channel and the right sound channel can advantageously be set to be a fixed negative number, such as -0.5m.

Preferably, the test signal comprises a maximum length sequence.

The correlation is usefully calculated in real time, in the time between samples of the test signal.

The reflection of said sound beam test signal is preferably detected at the listening position by an omnidirectional microphone. The invention provides in another aspect a method of outputting surround sound, said method comprising: performing the method described above; determining beam emission angles for a centre, left, right, surround left and surround right channel respectively; and emitting sound beams representing said respective channels in directions given by said respective angles.

The invention includes apparatus for outputting surround sound that is constructed and arranged to perform the method described.

In another aspect, the invention provides apparatus for outputting sound, said apparatus comprising: an array of output transducers; an input transducer; and a controller programmed: to cause said array to emit a test signal in the form of a directed sound beam into a room; to receive a signal from said input transducer that corresponds to at least one reflection of said sound beam test signal; and to determine a time of a maximum correlation between said emitted sound beam test signal and said received reflected sound beam test signal.

The input transducer is typically an omnidirectional microphone for placement at the listening position.

The controller is preferably arranged to calculate a path length for said directed sound beam.

The controller is preferably arranged to sweep said directed sound beam through a range of angles and to determine the magnitude of the maximum correlation between said emitted sound beam test signal and said reflected sound beam test signal for each angle in said range.

The apparatus preferably comprises a memory for storing, for each emission angle of said directed sound beam test signal, the time of said maximum correlation and preferably also the magnitude of said maximum correlation. The controller is preferably arranged to initially determine the position of said location in said room relative to said speaker system by receiving signals from said input transducer relating to direct, non-reflected, test signals.

The controller is preferably arranged to select a sound beam direction for a centre channel, which beam direction points from said speaker system directly towards said location in said room.

The controller is preferably arranged to select a sound beam direction for a centre channel, which beam direction is set to be the direction for which a maximum correlation exists between said emitted sound beam test signal and a directly received sound beam test signal by said input transducer.

The controller is preferably arranged to select an angle for at least one of the surround sound channels based on the determined time of maximum correlation.

The controller is preferably arranged to identify those angles for which a peak of maximum correlation exists.

The controller is preferably arranged to select a sound beam direction for a left sound channel, which direction corresponds to the angle at which there is the largest peak magnitude of maximum correlation out of all of the interrogated angles to the left of the centre channel.

The controller is preferably arranged to select a sound beam direction for a right sound channel, which direction corresponds to the angle at which there is the largest peak magnitude of maximum correlation out of all of the interrogated angles to the right of the centre channel.

The controller is preferably arranged to switch the speaker system to "direct mode" if the peak magnitude of maximum correlation for the left sound channel is smaller than a predetermined percentage of the peak magnitude of maximum correlation for the centre channel.

The controller is preferably arranged to switch the speaker system to "direct mode" if the peak magnitude of maximum correlation for the right sound channel is smaller than a predetermined percentage of the peak magnitude of maximum correlation for the centre channel.

The controller is preferably arranged to select a sound beam direction for the left surround sound channel, which direction corresponds to an angle at which there is a peak magnitude of maximum correlation to the left of the centre channel and which direction has a smaller beam angle and larger time of maximum correlation than the selected direction for the left sound channel.

The controller is preferably arranged to select a sound beam direction for the right surround sound channel, which direction corresponds to an angle at which there is a peak magnitude of maximum correlation to the right of the centre channel and which direction has a smaller beam angle and larger time of maximum correlation than the selected direction for the right sound channel.

The apparatus preferably further comprises a candidate beam determining means for determining a plurality of candidate beam directions for the surround sound channels and assigning a quality coefficient to each candidate.

The quality coefficient is preferably determined by multiplying the maximum correlation value for the beam angle by the beam path length.

The candidate beam determining means can be arranged to disregard a candidate beam under certain circumstances. This is done preferably by setting the quality coefficient to be zero for that beam. Any of the following conditions (a) to (e) may be used to disregard a beam:

(a) the magnitude of maximum correlation for the candidate beam direction is less than a predetermined percentage of the magnitude of maximum correlation for the beam direction of the centre channel;

(b) the candidate beam has a larger angle and a larger time of maximum correlation than an already selected left or right sound channel;

(c) the candidate beam has a smaller angle and a smaller time of maximum correlation than an already selected left or right sound channel;

(d) the candidate beam has a time of maximum correlation within, say, 200 μs and an emission angle within, say, 15° of an already determined candidate beam;

(e) the candidate beam is directed in a direction where the path length for adjacent angles differs by a predetermined amount from the candidate beam path length.

The controller is preferably arranged to select the candidate beam direction having the largest quality coefficient for a surround sound channel.

The controller is preferably arranged to set a "corner mode" for the speaker system if the angle between the identified left sound channel and identified right sound channel is smaller than a predetermined angle. In such cases, the controller is arranged to emit the left sound channel and right sound channel in "direct mode".

hi the "direct mode", the controller can be arranged to direct the left sound channel and right sound channel in the same direction as the centre sound and to use a subset of the transducers in the array to emit the left sound channel and right sound channel. Preferably, in addition, in direct mode, the controller can be arranged to set the focal length of the left sound channel and the right sound channel to be a fixed negative number, such as -0.5m.

The controller is preferably arranged to calculate the correlation in real time, in the time between samples of the test signal.

The invention will now be further described, by way of example only, with reference to the accompanying drawings, in which:-

Figure 1 shows a front view of a Sound Projector according to an embodiment of the invention comprising a two-dimensional array of output transducers; Figure 2 shows a front view of a Sound Projector according to another embodiment of the present invention, comprising a line array of output transducers;

Figure 3 schematically depicts exemplary signal paths for the Sound Projector of the present invention; Figure 4 shows a plan view of a room environment and three exemplary sound beams produced by the Sound Projector of the present invention;

Figure 5 shows a sonogram useful in explaining the present invention;

Figure 6 shows a plan view of a room environment and five exemplary sound beams that create a full Surround Sound effect; Figure 7 is a graph showing the maximum magnitude of the correlation between emitted and received test signals together with the corresponding time delay between emitting and receiving the test signals, for the room environment of Figure 6;

Figure 8 shows a plan view of a room environment and five exemplary sound beams that create a full Surround Sound effect; Figure 9 is a graph showing the maximum magnitude of the correlation between emitted and received test signals together with the corresponding time delay between emitting and receiving the test signals, for the room environment of Figure 8;

Figure 10 shows a plan view of a room environment and three exemplary sound beams when the Sound Projector is located in a corner; Figure 11 is a graph showing the maximum magnitude of the correlation between emitted and received test signals together with the corresponding time delay between emitting and receiving the test signals, for the room environment of Figure 10;

Figure 12 is a flow diagram showing how the beam directions for the left and right sound channels can be selected in accordance with the invention; Figure 13 is a flow diagram showing how quality coefficients can be determined for particular candidate beam directions in accordance with the present invention.

Figure 1 shows a typical Sound Projector system to which the present invention may be applied. The system 100 comprises a front panel 101 that is generally planar and rectangular and upon which 254 individual acoustic output transducers 111 are mounted in an array 110. As has been described in the documents referenced above, differently delayed versions of the same signal may be applied to the output transducers, so as to produce a sound beam that travels in a particular predetermined direction relative to the speaker system.

In the construction of Figure 1, the output transducers 111 extend both horizontally and vertically. Accordingly, this Sound Projector has a capability to steer sound beams in both the left/right and the up/down directions.

Figure 2 shows an alternative embodiment in which the casing 101 comprises 16 output transducers 111 arranged in a horizontal line. This embodiment is capable of directing beams only in the left/right direction. Although more limited, this construction is nevertheless well adapted for providing surround sound.

The exact number of transducers used in the Sound Projector is not critical, provided enough are present to ensure that the sound beams can be adequately directed in useful directions. Preferably, there are at least five transducers in the array 110. More preferably, there are at least 10 transducers, or more preferably still at least 16 transducers.

The transducers 111 are most preferably arranged such that sound beams can be directed at least in the left/right direction. However, embodiments (such as that shown in Figure 1) where sound beams are also capable of being directed in the up/down direction are included within the scope of the invention.

Figure 3 shows a simplified block diagram of the signal path through the Sound Projector. An input audio signal 190 is fed to a decoder 195. The audio signal may be a Dolby Digital 5.1 audio signal from a DVD, television decoder, Blu-Ray disk or any other type of audio signal source. The decoder 195 may be any type of conventional decoder, such as a Dolby 5.1 decoder. The decoder 195 outputs in this embodiment five separate audio signals, 200, 202, 204, 206, and 208. In this embodiment, the audio signals each correspond to a respective channel in a Dolby 5.1 system, namely left 200, right 202, centre 204, left surround 206 and right surround 208. The subwoofer channel is not shown although this channel may usually also be decoded by the decoder 195 and sent to a conventional subwoofer for output. Each channel signal 200, 202, 204, 206, 208 is provided to a separate delay element 210 that imposes a channel-specific delay on the signal. This channel-specific delay will be related to the distance that sound waves for that channel must travel before they reach the user. In general, longer delays will be provided for channels that travel a shorter distance through the air. This helps to ensure that each channel reaches the listening position (where the user is) substantially simultaneously.

Each delayed channel is then preferably fed to a replication and delay means 220. The replication and delay means 220 replicates (that is to say copies) the input signal a certain number of times (five times in Figure 3) and provides a selected delay to each replica input signal. The delay is selected in accordance with the direction in which the sound beam for that channel is to be directed. In this embodiment, the delay is also selected bearing in mind the position of the particular output transducers 250.

In Figure 3, each replication and delay means 220 produces five individually delayed replicas, one for each of the output transducers 250. Accordingly, in this embodiment, all of the output transducers 250 are used to output each of the channels 200, 202, 204, 206 and 208. hi some other embodiments, it is possible to use a subset of the output transducers 250 to output certain of the channels and not all transducers 250 need to be used to output all of the channels. In general, there will be more than five output transducers and in such cases, each replication and delay means 220 can have more than five outputs.

Each output transducer 250 has associated with it an amplifier 240 and a summing means 230 which receives a plurality of delayed replicas from the replication and delay means 220. In this embodiment, each summing means 230 receives a replica from each of the replication and delay means 220. Accordingly, each summing means 230 in this embodiment outputs the sum of five different signals. In general, the summing means 230 for a particular output transducer 250 will receive a signal from a replication and delay means 220 in accordance with the particular choice of transducers to be used to output a particular channel. Figure 4 shows a plan view room of a room 300 in which a Sound Projector 100 has been placed. The position marked "X" is the listening position for the user. In an ideal configuration, the user will receive five channels, each corresponding to one of the surround sound channels: left, right, centre, surround left and surround right. Ideally, the sound will reach the listener from the intended direction for each channel.

The arrow marked Bl in Figure 4 shows the centre channel path. This channel can be created by energising each transducer in the array with a signal corresponding to the centre channel (with no relative delay between any of the transducers). In this case, the replication and delay means 220 for the centre channel merely replicates the signal and does not apply any relative delay amongst the signals for each of the output transducers 250. Alternatively, a sound beam can be produced and the sound beam can be focussed, as described in the above mentioned prior art documents. When a focussed sound beam is output, there will be a relative delay amongst the transducers for the centre channel. Generally, transducers near the middle of the array will be delayed more than transducers near the edge to provide a focussed beam. A negative focus can be provided, which creates a more widely dispersed sound beam.

The arrow marked B2 in Figure 4 shows a typical path for a right surround sound channel. As can be seen, this sound beam is reflected from the right sidewall and the back wall so as to arrive at the listening position from behind. Sound for this channel travels the longest distance.

The arrow referenced B3 in Figure 4 shows a typical sound path for the left channel. As can be seen, the sound beam is reflected once from the left sidewall to arrive at the listening position substantially from a left-hand direction.

Although Figure 4 shows only 3 sound beams, in most practical embodiments of the Sound Projector there will be five sound beams. As well as the left channel sound beam B3, there will also be a mirror-image right channel sound beam and as well as the right surround sound beam B2, there will be a mirror image left surround sound beam. Accordingly, a full surround sound experience can be provided for the user using a single device 100.

With reference to Figure 3, the delay means 210 will typically apply the longest delay to the centre channel 200 (represented by Bl in Figure 4) because this channel travels the shortest distance. A medium-length delay can be applied to the left and right channels 202, 204 (represented by B3 in Figure 4) because these channels travel the second shortest distance. No delay (or perhaps a very small delay) can be applied to the surround sound channels 206, 208 (represented by B2 in Figure 4) because these channels usually travel the longest distance. The delays are preferably chosen such that temporally corresponding portions of the various channels reach the listening position "X" at the same time.

Where the surround sound is provided in combination with a video display, an appropriate delay may be applied to the video signal to ensure that the video displayed also corresponds temporally with the sound reaching the user at any point in time.

From Figure 4 it can be noted that the left and right channels are directed through a larger angle (as measured from a line perpendicularly extending from the front of the Sound Projector) than the surround sound channels. Also, the path length for the left and right channels is in Figure 4 shorter than the path length for the surround sound channels. Ln general, an optimum set-up for the system will be when the left and right channels each have a shorter path length and wider angle than the surround sound channels.

A set-up method according to the present invention will now be described.

Microphone Location

When setting up the system, the user is instructed to hold the microphone at the listening position in the room, preferably at the position where the user's head will be positioned during use of the Sound Projector. An optional first step for the set-up system includes locating the position of the microphone in the room relative to the Sound Projector.

In one simple embodiment, the user may simply input the location of the microphone to the Sound Projector, perhaps by inputting the distance from the Sound Projector to the microphone and perhaps also the angle of the microphone relative to the Sound Projector.

hi a more advanced embodiment, the Sound Projector may locate the microphone by triangulation, as disclosed in WO 01/23104. In such a method, at least three of the output transducers are used to output three test signals and the time taken for the test signals to arrive at the microphone from the Sound Projector is measured. The recorded times can be used to calculate the distance between each output transducer and the microphone and, from this, the position of the microphone relative to the Sound Projector can be determined. The test signals may be output simultaneously, in which case it is preferable that they are distinguishable from one another (for example by being located in different frequency ranges or by using different pseudorandom noise or MLS codes). Alternatively, the test signals can be emitted sequentially.

In a preferred embodiment, the following method is used to determine the location of the microphone, which is preferably omnidirectional.

The Sound Projector is preferably programmed to first output a sound beam in the straight- ahead direction (i.e. perpendicular to the major plane of the Sound Projector) and with a focal length of -0.5 m. Setting the focal length to a negative value gives a broadly dispersing beam. The signal output preferably corresponds to a maximum length sequence (MLS) signal. The maximum length sequence test signal has a length of N samples. Preferably, 1023 samples are used. This gives a good balance between being able to well correlate the MLS test signal and performing the method in a short period of time. The sound received at the microphone is sampled, preferably at the same sampling rate as the MLS sequence, and correlated with the MLS signal. The correlation is preferably performed in real time every sample period between the last N samples received at the microphone and the MLS test signal. Preferably, such correlation is carried out in real time every sample period, until enough time has passed for sound that has travelled a maximum expected direct microphone-Sound Projector distance to reach the microphone. This maximum direct microphone distance can be set to, say, 6 m or 7 m. It can be user definable if desired. For example, when the maximum length sequence has a length of 1023 bits and when the sample rate is 48 kHz, and assuming a speed of sound of 340 m/s, a distance of 7 m corresponds to a time for sound to travel that distance of about 20.6 ms which corresponds to 988 sample periods. Accordingly, the sampled signal from the microphone is correlated to the original MLS test signal at every sample period between 1023 and 2011 samples. After this, 988 values of the correlation, each corresponding to a slightly different moment in time will be available.

Preferably, the value of this correlation that is highest is identified and the system can thereby determine that this value corresponds to the reception by the microphone of the MLS signal as it travels directly from the Sound Projector to the microphone. In general, reflected signals will have a lower correlation due to their increased attenuation relative to the direct signal. The point in time of this maximum correlation can be determined and used to determine the distance (direct path length) from the Sound Projector to the microphone. For example, if the maximum correlation occurs after 1500 sample periods, then this means that the sound has taken 477 sample periods to travel from the Sound Projector to the microphone. This corresponds to a time delay of 9.9 ms, which in turn corresponds to a distance of 3.37 m. Accordingly, under such a circumstance, the direct distance between the Sound Projector and the microphone would be determined as 3.37m.

There will inevitably be some part of the measured time delay (i.e. the time between emitting the MLS and detecting the maximum correlation at the microphone) that corresponds to delays through the system electronics processing. This system delay can be quantified and represented as a constant time offset. This time offset can then be subtracted from the measured delay values to give a more accurate time of flight measurement for the audio signal.

The relative angle between the microphone and the Sound Projector can then be determined, hi this document, all angles are expressed relative to a line extending perpendicularly out from the centre of the Sound Projector. This perpendicular line is denoted by an angle of 0°. Sound beams that travel to right of this line (as viewed by the user) are denoted by positive angles and sound beams that travel to the left of this line (as viewed by the user) are given negative angles. A full sweep of the beam through the half space in front of the Sound Projector from left to right will therefore be denoted as a sweep from -90° to +90°. In order to determine the angle of the microphone, the Sound Projector is arranged to repetitively output a sound beam in a sweep from the left to the right. This sweep may encompass the full half plane of -90° to +90°, but preferably covers a smaller range of angles than this because it is unlikely that the microphone will be located at extreme angles relative to the Sound Projector. In the preferred embodiment, the sweep is from -76° to +76°.

The focal length of the sound beam used is set to be the same as the distance between the microphone and the Sound Projector determined in the first stage. Accordingly, when a beam having an angle that directly points to the microphone is output, it will be focussed on the microphone and the microphone will receive an intense signal. Once again, an MLS test signal having N samples is output as a sound beam directed at a desired angle (for example the first beam will be output at -76°). The MLS test signal is followed by silence. Once again, the signal received at the microphone is correlated with the original MLS signal every sample period, until the maximum expected direct microphone distance is reached. This maximum expected direct microphone distance can, once again, be set to 6 m or 7 m, although it is preferably related to the distance determined in the first stage, for example by multiplying that distance by 1.5 to give a margin of error. Taking as an example the case when the microphone is determined to be 3.37 m from the Sound Projector in the first stage, an expected microphone distance of no more than 5 m (1.5 x 3.37) may be used in the second stage. This corresponds to 706 sample periods. Accordingly, the correlation between the emitted sound beam test signal and the received sound beam test signal will be performed for each of 706 sample periods following the emission of the last bit of the MLS.

According to the invention, the time of maximum correlation is determined. This can be determined as a number of samples, as a period of time in seconds or as a sound beam path length. Preferably, the magnitude of the maximum correlation is also stored. This is repeated for each angle within the swept range of interest. Typically, the angles are incremented in one degree steps from -76° to +76°. Once this sweep has been completed, the speaker system will be in possession of the magnitude of maximum correlation and the corresponding time (or, equivalently, path length or number of samples) for each angle within the swept range of angles. The system can then determine that the angle corresponding to the overall largest magnitude of maximum correlation is the angle of the microphone relative to the Sound Projector. As a check, the distance between the Sound Projector and microphone is preferably calculated again at this point and compared with that obtained or input previously.

Room Scanning

Once the microphone position has been determined (by any of the above methods), the next stage of the set-up procedure can be performed. In this stage, a sweep is once again performed through all angles of interest, although this time with a view to receiving only reflections of the sound beam. Accordingly, a maximum expected travelling distance for the sound beam is set to be much greater, typically 3, 4, 5, 6 or 7 times greater than the distance from the Sound Projector to the microphone. The system may assume a sufficiently high fixed number, for example 24 m, for the expected travelling distance of sound beams from the Sound Projector to the microphone, which will in general travel via a reflection from one or more of the walls/ceiling/floor etc.

Once again, a beam is output at each angle within the range of angles, for example -76° to +76°. Again, an MLS test signal is used and correlated with the received microphone signal every sample period. The correlation continues for enough time to account for all signal paths of interest, for example enough time to account for 24 m of sound beam propagation at each angle. This corresponds to approximately 70 ms or 3300 sample periods (at a sampling frequency of 48kHz) at each angle.

For this sweep, the focal length of the beam is set to be 1-2 times the distance from the Sound Projector to the microphone, for example 1.3 to 1.8 times, more preferably 1.5 times this distance. Accordingly, the microphone will receive maximum power when receiving a reflected signal that travels a distance roughly corresponding to the focal length. When a sound beam is output from the Sound Projector, alias beams will in general be created (otherwise known as side lobes). These alias beams are generally stronger for larger deflection angles than smaller deflection angles. In some cases, the side lobe energy received at the microphone direct from the Sound Projector can significantly exceed the energy of the useful reflected beam incident on the microphone. Accordingly, when a beam is directed at an angle other than 0°, the possibility remains that alias beams will reach the microphone first. Thus, a time mask can be applied to signals that are directed in a range outside of a small 20° window around the centre channel. No time mask will be applied for signals directed straight at the microphone and for signals directed 10° either side of the microphone. However, for signals directed at angles outside of this range, the correlation calculation is preferably started only after a certain time has passed. This time will preferably be chosen such that it exceeds the time that it would take for signals to travel directly from the Sound Projector to the microphone. A margin of error can be implemented to ensure that no direct signals at all are received by the microphone when beams are directed outside of the central window. Accordingly, if the microphone is

3.37 m from the Sound Projector, the system can be programmed to disregard all signals having a travelling distance of 4 m or less that are directed outside the 20° zone pointing at the microphone. This ensures that alias signals travelling along a direct path from the Sound Projector to the microphone are not taken into account. This time mask can be represented as:-

m(θ) = dc + 0.7 for |θ-θc|>10° m(θ) = 0 for|θ-θc| < 10°

Where θc is the angle of the microphone, Cl0 is the distance to the microphone in metres. m(θ) is expressed in metres. So for angles outside of a small 20° window around the centre channel, correlations are only performed for distances 0.7 m greater than the direct path length.

Correlations are performed every sample period between the emitted MLS test signal and the reflected test signal received at the microphone. This is repeated for every angle within the swept range. The above processing is preferably all carried out in a real time at a sample rate of 48 KHz. This can be achieved by using a typical low-cost audio signal processing device.

Figure 5 shows a sonogram summarising the signals detectable at the microphone. The swept angles are shown on the x-axis, the distance (path length) is shown on the y-axis and the value of the correlation (whereby a correlation of 0 dB corresponds to a perfect correlation) is denoted by the intensity of shading in the Figure. Correlations that are only partial have negative values measured in dB in Figure 5. Elsewhere, where a correlation is given as a positive value, a correlation value of "1 " corresponds to 0 dB in Figure 5. In this sonogram, only path lengths greater than 2.8 m and smaller than about 11.3 m are shown. Several peaks of correlation can be identified as dark points on the sonogram.

It can be seen from Figure 5 that a well-correlated signal is received by the microphone at an angle of 0°, with path length of approximately 3.2 m. This is denoted on the sonogram by the reference numeral 10. This is the signal corresponding to the direct travel of the sound beam from the speaker system to the microphone. Reference numeral 12 shows a slightly longer path length and this peak of correlation is created by sound beams reflecting once from the floor before reaching the microphone. Reference numeral 14 shows a longer path length still and this relates to sound beams that have reflected once from the ceiling before reaching the microphone. Reference numeral 20 is a yet longer path length and this could relate to a signal that has reflected from both the floor and the ceiling or from some piece of furniture such as the seat. Strong peaks of correlation are shown at angles of approximately ±50° (referenced 16 and 18 in Figure 5) and these relate to sound beams that have bounced once from the left and right walls respectively. Smaller peaks are shown having slightly longer path lengths and similar angles (referenced 22, 24, 26 and 28 in

Figure 5) and these probably represent sound beams that have bounced from the ceiling or floor before hitting left and right walls. Numeral 30 identifies the correlation obtained from a sound beam that has travelled to the back of the room and directly bounced back from the rear wall. Correlations 32 and 34 are at angles of approximately ±30° and have a path length of 9.2 m, indicating that these correlations are for sound beams that have bounced once from the left/right wall and once on the rear wall (see the beam B2 in Figure 4). Correlations at positions 36 and 38 are at wide angles of ±65° and are likely to be triple bounces from the left and right walls.

The following method is utilised to, preferably automatically, place the 5 audio channels along particular beam paths that will provide an optimum surround sound effect in a robust manner. As can be seen from Figure 5, a great many potential beam locations exist and it is necessary to disregard certain candidate beam locations based on certain predetermined criteria.

Figures 6 and 7 show an example that will be explained in greater detail. Figure 6 shows a plan view of a typical listening room 300, in which the microphone is placed on a seat located 4.09 m from the front wall of the room, and where the sound projector 100 is placed centrally on the front wall 302. The desired sound beams to achieve a surround sound effect are illustrated in Figure 6.

Figure 7 shows two graphs. The darker line illustrates the magnitude of maximum correlation for each angle in the range -76° to 76°. The grey line shows the time that that maximum correlation took place. This time is directly related to the path length that the sound beam takes. Preferably, the Sound Projector stores in memory a value for the maximum correlation and a value for the corresponding time in respect of each angle in the swept range. This limits the amount of data that needs to be stored to only two data words per angle. If the entire sonogram of Figure 5 were stored, this would involve storing over 500,000 data words. The Micronas MSP-M device does not have enough memory to store the entire sonogram whereas it does have enough memory to store the data of Figure 7.

It has been found by the present inventors that the data of Figure 7 (i.e. the smoothed magnitude of maximum correlation data for each angle and the time data (or path length data) for each angle) can be used to automatically attribute the sound channels to particular directions in a very satisfactory and robust manner. The darker line in Figure 7 is related to the sonogram in Figure 5. For each angle in Figure 5, the maximum value of correlation (wherever it occurs) is identified and stored. At the same time, the distance (or, equivalently, the time) at which this maximum value of correlation occurs is also stored. A graph of the maximum correlation with respect to angle can then be drawn (see the darker line in Figure 7). Similarly, a graph of the distance that the sound beam travels to achieve this maximum correlation can also be drawn (see the grey line in Figure 7). Note that the data of Figures 7, 9 and 11 have been normalised but this is not necessary to carry out the invention.

Due to noise, the data may be somewhat jerky. This can be overcome by performing thee abovementioned sweep of the sound beam more than once, for example two to five times, and taking an average of values to obtain the graph of Figure 7. Another possibility is simply to smooth the data and this is what has been done to provide the graph of Figure 7. The maximum correlation amplitude data C(n) is smoothed by using a simple mean of three consecutive elements. So:

C1(Ii) = [C(n-1)+C(n)+C(n+1)] / 3 for n = -74°...+74°.

where Cs is the smoothed data vector, hi this case, the first and last elements of the correlation amplitude data (C(-75) and C(+75)) are just copied.

The time data (or equivalently path length data) is not usually smoothed.

Post-processing

The result of the above real-time processing is that the speaker system memory has stored therein two items of data associated with each angle in the swept range. This can be stored as two vectors, one for the correlation data and one for the time data. As explained above, this data can be obtained in real time in a matter of seconds or minutes using the Micronas MSP platform.

The data can then be manipulated in a post-processing procedure to determine where to place the sound channels. Preferably, an algorithm is used to choose the beam angles for each of the audio channels. Parts of the algorithm are summarised in Figures 12 and 13. The algorithm needs to be robust to account for cases where the Sound Projector is placed in a poor listening environment or in a position that makes surround sound reproduction difficult (such as in the comer of a room).

The first step in the procedure is to place the centre channel. This will usually be placed along the direct path from the speaker system to the microphone and will correspond to the angle for which the maximum correlation is the largest across the whole range of swept angles, hi Figures 6 and 7, the centre channel is directed at approximately +2.5°.

The next stage is to identify the best beam candidates lying to the left and to the right of the centre channel respectively. Firstly, all the potential candidate beams are determined. This is achieved by determining all of the angles for which a peak of maximum correlation exists. A peak is defined as existing at an angle where the maximum correlation value at that angle is larger than the correlation values lying at the two neighbouring angles. The already determined centre channel is not considered to be a "peak" in this analysis and, preferably, any peak located within 12° of the centre channel are not included. Preferably, all of the peaks present (except those near the centre channel) are identified although it may save on processing if a finite number of the largest peaks are identified, for example 6 peaks on either side of the centre channel, hi Figure 7, six peaks have been identified each side of the centre channel. These candidate beams are numbered 50-61 respectively as shown in Figure 7. The candidate beams are then ranked in order of maximum correlation value. Two lists of candidate beams are thus obtained, one for the left and one for the right, hi the example of Figure 7, the ranking is as follows for the left side:

51, 54, 55, 52, 53, 50

For the right side of Figure 7 the ranking is as follows:

59, 56, 61, 60, 58, 57

The two largest peaks of correlation lying to the left and right of the centre beam are identified. From the above lists, these can be seen to be candidate beam 51 on the left and candidate beam 59 on the right. The system then goes about selecting the candidate beam most suitable for the left sound channel and the right sound channel. This selection process is summarised in the flow chart of Figure 12.

As shown in Figure 12, the system starts by looking at the angle subtended between the two largest candidate beams. Candidate beam 51 lies at an angle of -45° whereas candidate beam 59 lies at an angle of 47°. The angle subtended between the two beams is thus 92°. It is then determined whether this subtended angle is greater than or less than a predetermined threshold angle. Good results have been found when this predetermined threshold angle is set to be 60°, although other angles such as 50° or 70° could be used. If the angle between the two candidate beams having the largest peak correlation is smaller than the threshold (which is not the case in Figures 7 or 9 but is the case in Figure 11), then the system determines that the Sound Projector is probably located in the corner of a room. A flag is then set to indicate that the Sound Projector is to be operated in "corner mode".

This identification of whether the Sound Projector is located in a corner is a very useful function, because otherwise it would be necessary to request that the user indicates whether the Sound Projector is in the corner. Accordingly, this facility allows the set-up procedure to be conducted with a high degree of autonomy.

If, as shown in Figure 7, the angle between the two largest peaks of correlation is greater than the predetermined value (e.g. greater than 60°) then the angles having the two largest peaks of correlation are determined to be the angles for the left sound channel and right sound channel respectively. Accordingly, the Sound Projector would at this stage allocate the left sound channel (referenced B 3 in Figure 4) to the candidate beam 51 and allocate the right sound channel to the candidate beam 59.

A check is then made to determine whether a more suitable candidate beam exists for the left and right channels. This is described in detail in Figure 12. A beam is considered to be a more suitable candidate if each of the following conditions exist:

(a) the candidate beam has a maximum correlation value that is at least 10% of the correlation value for the direct sound beam from the Sound Projector to the microphone; and (b) the candidate beam has a larger angle than the already selected beam for the left or right sound channel; and (c) the candidate beam has a maximum correlation value that is within 3 db of the maximum correlation value for the already selected left or right sound channel; and (d) the candidate beam has a shorter path length than the beam already selected for the left or right channel

If each of these conditions are met, then that candidate beam is instead chosen for the left or right sound channel respectively and the beam previously chosen (which has the highest correlation) is put at the top of the list of potential candidates for the surround sound channels.

The purpose of this checking procedure is to ensure that, in the case when a surround sound beam is the strongest, this is not accidentally chosen for the left or right sound channels. In general, as shown in Figure 4, the left or right sound channels will have a greater angle than the surround sound channels and will have comparable or greater strengths than the surround sound channels. Also, they will in general have a shorter path length than the surround sound channels. The four questions above (as shown in Figure 12) ensure that these conditions are met.

It then becomes necessary to select the candidate beam to use for the surround sound channels. This can be done by discarding the least suitable beams. Preferably, this is achieved by calculating a quality coefficient for each candidate beam in the list. One method for carrying this out is shown in Figure 13.

Initially, the quality coefficient for a candidate beam is made to be equal to the value of the correlation multiplied by the path length for that beam. However, to ensure that beams travelling very long path lengths, which are likely to be due to triple or quadruple bounces, do not dominate, the path length is limited at two times the path length chosen for the left or right channel beams, respectively.

Taking Figure 7 as an example, the left channel beam has already been selected as candidate beam 51. The path length for this candidate is shown in Figure 7 as having a relative magnitude of 0.31. This path length is multiplied by two to obtain the maximum path length (0.62) that will be used when calculating the quality coefficient for the other beams.

Referring briefly to Figure 9, if the candidate beam 66 was chosen to be the left channel beam, it can be seen that this has a path length of 0.29 and so the maximum path length used in calculating the quality coefficient would be 0.58. Thus, when calculating the quality coefficient for the candidate beam 62, a path length of 0.58 would be used, even though this beam has an actual path length of 0.72.

Returning to Figure 7, the quality coefficient for each of the beams can be calculated. The following values are obtained for each of the beams on the left side.

Candidate beam Ouahtv coefficient

50 0.21 x 0.5 0.105

52 0.27 x 0.42 0.1134

53 0.26 x 0.49 0.1274

54 0.36 x 0.49 0.1764

55 0.28 x 0.57 0.1596

This assignation of quality coefficients already gives a rough idea of which beams are suitable for the surround sound channel. However, as shown in Figure 6, it is preferable that the surround sound channel has a smaller angle and a longer path length than the corresponding left or right sound channel. Also, it is preferable that the surround sound channel is not too weak. Accordingly, Figure 13 shows a series of conditions that the candidate beam must meet in order to be considered for the surround sound channel. If the candidate beam does not meet each of the conditions then it is disregarded. The candidate beam is preferably disregarded in practice by setting its quality coefficient to be zero. In the first question of Figure 13, it is determined whether the maximum value of correlation for that candidate beam is smaller than some percentage of the channel centre correlation. In Figure 7, the centre correlation is 1. A threshold value of 7.5% can be selected, although other thresholds may be used, such as 5%, 10%, 15% etc. In Figure 7, it is the case that every candidate beam has a correlation that is at least 7.5% of the correlation for the centre beam (i.e. the correlation is at least 0.075). This test is designed to disregard weak reflections. A normalised threshold may instead be used, for example by determining (C - Cmin)/( Ccentre - Cmin) < 0.05 where C is the amplitude of the maximum correlation for the candidate beam, Ccentre is the amplitude of the maximum correlation for the centre channel beam and Cmin is the smallest maximum correlation from all of the angles.

Next it is determined whether both the angle and path length are greater than the angle and path length for the respective left or right beam. If the angle and path length are both greater, then this can indicate a double bounce off the left or right walls or even a triple or quadruple bounce. It can be seen from Figure 7 that candidate beams 50 and 61 have a greater angle and longer path length than the left channel beams 51 and 59 respectively. Accordingly, the quality coefficient for the candidate beams 50 and 61 is set to be zero.

Next, it is determined whether both the angle and the path length is smaller than the left or right sound channels, respectively. If the angle is smaller and the path length is smaller, then this could indicate that the beam is taking a more direct route, for example by being bounced from the floor or ceiling before reaching the microphone. None of the beams of Figure 7 are disregarded due to this test.

Next, it is determined whether the beam is very close to any of the other candidates higher in the list. In particular, it is determined whether the candidate beam angle is within a certain angle range from a previously considered candidate beam and it is determined whether the path length is within a certain path length range from that other candidate beam. If two candidate beams are within 15° of one another and have a path length within ten samples of one another, then it is determined that the two candidates are actually part of one and the same beam and the quality coefficient for the candidate having the lower correlation is set to be zero. In Figure 7, candidate beam 53 is within 15° of already considered candidate beam 52 and has a path length within ten samples. Accordingly, the quality coefficient for candidate beam 53 is set to zero. The threshold values of 15° and ten samples may of course be varied in accordance with the circumstances. . Preferably the threshold value for angle is selected from the range 5° to 25°. Also, instead of measuring the threshold in samples, it may be measured in time (seconds) or distance (metres). Samples, time and distance are all interchangeable in this respect. Preferably the threshold value is selected such as to detect time differences between path lengths in the range 100-500 μs.

Next it is determined whether the candidate beams are "robust". If the candidate beams lie at an angle where there are great changes in path length for angles very close to the candidate beams, then this tends to indicate that the candidate beams are quite unstable, perhaps because they are directed at some piece of furniture having an unusual shape. It is firstly checked whether the path length for the angles either side of the candidate beam is within 20% of the candidate beam path length. So, for example, taking candidate beam 54 in Figure 7, it can be seen that this candidate beam lies at an angle of -27°. The path length here is 0.49. This path length is compared to the path length at -26° and -28°. In this example, these path lengths are also 0.49. When the adjacent path lengths are within a certain threshold of the candidate beam path length, for example within 20%, then it is determined that the candidate beam is a good candidate. Candidate beam 54 in Figure 7 passes this test.

If it has been determined that the path lengths either side of the candidate beam do not fall within 20% of the candidate beam path length, then a further test is carried out. hi more particularity, this test determines whether the path length for each of the two angles either side of the candidate beam is within ten samples of the candidate path length. So, if the candidate beam lies at +40°, the path lengths at +38°, +39°, +41 ° and +42° are checked. If all four path lengths are within 10 samples of the candidate beam path length, the candidate beam passes the test. Accordingly, this test can be seen as determining whether the path length fulfils either of the following two conditions: (a) the path length for the two angles that are each directly either side of the candidate beam is within 20% of the candidate beam path length; or

(b) the path lengths for the four nearest angles to the candidate beam are each within ten samples of the candidate beam path length

If either condition is fulfilled, the candidate beam is not disregarded due to this test. If both conditions are not fulfilled, then the candidate beam fails the test and the quality coefficient for that candidate beam is set to zero.

Once these tests have been carried out (for both the left-side and the right-side separately - see Figure 13), the left and right surround sound channels are selected to be directed at an angle corresponding to the candidate beam having the highest quality coefficient for that side, hi Figure 7, the left surround sound channel will accordingly be directed at -27° (candidate beam 54) and the right surround sound channel will be directed at 24° (candidate beam 56).

hi the event that all of the quality coefficients are zero, the surround sound channels can be steered at the same angle as the respective left or right sound channels. Accordingly, if on the left side there are no candidate beams having a non-zero quality coefficient, the left surround sound channel will be steered in the same direction as the left sound channel.

A final check is done once the left sound and left surround sound channels have been selected. If the angle of the left surround sound channel is larger than the angle of the left sound channel, the two channels are swapped. Similarly, if the angle of the right surround sound channel is larger than the angle of the right sound channel, these two channels are swapped. This ensures that the left and right sound channels have a greater angle than the left and right surround sound channels, respectively.

Figures 8 and 9 show another example where the sound projector is not centrally located in the listening room, hi this case, as shown in Figure 8, sound beams directed to the right generally have a longer path length than sound beams directed to the left. As shown in Figure 8, this also tends to mean that the channels directed to the right will have wider angles than the channels directed to the left. These facts are represented in Figure 9.

Due to the fact that they are the largest peaks in Figure 9 (other than the centre channel), candidate beams 66 and 71 are initially selected as the left sound channel and right sound channel respectively. No other beam meets the requirements of Figure 12 and so these beams remain the left and right sound channels even after the process of Figure 12 has been carried out.

It is then necessary to select the surround sound channels. Candidate beams 62, 63 and 64 have a longer path length and wider angle than candidate beam 66 and so are each disregarded. Similarly, candidate beams 72 and 73 are also disregarded for the same reason. These candidate beams likely relate to double or triple bounces from the left and right sidewalls. Candidate beam 65 is disregarded as having the same path length and being within 15° of candidate beam 66. Candidate beam 67 is the only remaining beam on the left-hand side having a non zero quality coefficient and thus is selected for the left surround channel.

For the right surround channel, there is a choice of candidate beam 68, 69 or 70. The quality coefficients are as follows:

68: 0.21 x 0.6 = 0.126

69: 0.24 x 0.5 = 0.12

70: 0.27 x 0.45 = 0.1215

The three candidate beams 68, 69, 70 each have significantly different path lengths to one another, and the path length is in all three cases greater than the path length for the selected right channel. Furthermore, the three beams are each directed at an angle that is quite robust, because the path length for angles either side of the candidate beams are quite similar. Accordingly, the candidate beam having the largest quality coefficient will be selected, which in this case is beam 68.

Figures 10 and 11 show an example where the Sound Projector is located in a corner of a room. The largest peaks of maximum correlation either side of the centre beam are at -9° (beam 79) and 25 ° (beam 80). Accordingly, the angle between these two candidate beams (34°) is smaller than the predetermined threshold (60°) and thus the Sound Projector determines that it has been located in the corner of the room. It accordingly automatically switches its mode to "corner mode".

hi "corner mode" reflected sound beams are used only for the surround sound channels whereas the left and right sound channels are not reflected. Instead, the left and right sound channels are directed in the same direction as the centre channel, preferably using a subset of the transducers located on the left-side or right-side of the Sound Projector. This is called "direct mode". For example, in the case where the Sound Projector is configured as shown in Figure 2, the leftmost six transducers can be used to output the left sound channel and the rightmost six transducers can be used to output the right sound channel.

The method of Figure 13 is then be used to determine the candidate beams for the surround sound channels. The result of this method is that candidate beams 79 and 80 are used for the left surround sound and right surround sound channel respectively.

The left and right sound channels may also be output directly, even when it is determined that the Sound Projector is not located in the corner of the room, hi this so-called "direct mode", the left and right sound channels are output directly by subsets of the transducers, in the manner shown in Figure 10. This "direct mode" will generally be used when all of the beam candidates are very weak (for example when all of the beam candidates have a maximum correlation magnitude smaller than 10% of the correlation magnitude for the centre channel). In such cases, it is better to use the reflected beams for the surround sound channels and operate the (more important) left and right sound channels in "direct mode".

In "direct mode" the focal length for each of the left sound channel and the right sound channel is set to be -0.5 m (to create a diffuse sound field) and the steering angle is set equal to the centre channel angle. Preferably, different subsets of transducers are used for the left sound channel and the right sound channel. As described above, the six left-most transducers can be used for the left sound channel and the six right-most transducers can be used for the right sound channel. As discussed, "direct mode" is used when the system identifies that it has been placed in the comer of the room or when the correlation for all of the candidate beams is very small. In direct mode, the candidate beam nevertheless having the largest correlation is chosen for the two surround sound channels and the focal length of each is set to be one half of the corresponding path length.

During playback of surround sound (i.e. when the Sound Projector is used by the listener following the set-up method), the centre channel is set to have a focal length of -0.5 m and the other channels are set to have a focal length corresponding to one half of their path length. This ensures a reasonable degree of dispersion at the listening position.

System Implementation

The system can be implemented using a Micronas MSP-M processor working on a sample by sample basis. A sampling rate of 48 kHz can be used because this is consistent with the usual sample rate that is used for audio reproduction. This gives a sample period of roughly 20.5 ms. hi the best mode for carrying out the invention, the processing relating to the correlation calculation is carried out during this sample period (i.e. in the time between samples). Also during each sample period, the necessary computation for forming a sound beam is carried out for the next output audio sample and the signals are generated to drive the transducers of the array.

During the set-up method, only one sound beam is needed and this saves on computational intensity. The Micronas platform has a IK analysis buffer and this can be used to receive the audio sample from the microphone in real time to allow the correlation to be carried out.

The system is advantageously configured as a state machine working on a sample basis although other configurations, for example by working on a block basis, are possible. The state machine routines for the microphone based audio set-up (MBAS) system are included below. MB ASJnitialise:

Set SteeringAngle to -76°.

Set PassCounter to 1. (Indicates the first pass of the process) Set FocalLength to 3 metres. Set OutputSample to zero, (output muted until MLS sequence begins)

Move state machine onto MBAS_IncrementBeamAngle state. Return.

MBAS_IncrementBeamAngle Increment SteeringAngle.

Check if SteeringAngle > +75°.

If so and this is the first pass, move state machine onto MBAS FirstPassComplete. If so and this is the second pass, move state machine onto MBAS_SecondPassComplete. Otherwise move state machine onto MB AS_SteerBeam.

Return.

MBAS_SteerBeam Calculate the beam forming coefficients for the given angle and focal length.

Set MBAS_Counter to 1023. Move state machine onto MBAS_GenerateMLS. Return.

MBAS_GenerateMLS

Decrement the MBAS_Counter. If it reaches zero:

If this is the first pass move state machine onto MBAS_AnaliseMLS. If this is the second pass move state machine onto MBAS_DoTimeMask. Return.

Else:

Generate the next sample of the MLS sequence and write it to the OutputSample register.

Also store it in the IK MLS sequence buffer. (MLS generation algorithm is described in appendix A.) Return.

MBAS_DoTimeMask

Increment the MBAS_Counter.

If (|SteeringAngle-mic angle|>10) AND (MBAS_Counter < micPathLength+1) Then stay in MBAS_DoTimeMake state. Else move onto MBAS AnaliseMLS state.

Return.

MBAS_AnaliseMLS

Perform correlation of the analysis buffer contents with the MLS sequence. Take the absolute value of the correlation.

Peak hold the biggest correlation so far and remember the corresponding delay (MBAS_Counter).

Increment the MBAS Counter. (effectively the distance represented in sample periods) Check if maximum distance has been reached. (6m on first pass, 24m on second).

If so, move onto MB AS_CorrelationsDone. Otherwise, remain in MBAS_AnaliseMLS state. Return.

MBAS_CorrelationDone

Store the peak held correlation amplitude for the present steering angle.

Store the corresponding delay for the present steering angle.

Move back to MBAS_IncrementBeamAngle state.

Return.

MBAS_FirstPassComplete

Find the biggest of the peak correlations. This corresponds to the mic position. Also record the corresponding path length to the mic. Set PassCounter to 2. (as we move onto the second pass) Set SteeringAngle to -76°.

Set the focal length to 1.5 times the mic path length. If the mic angle >30° or the correlation < 0.005 then report an error and exit MBAS.

Move back to MBAS IncrementBeamAngle state. Return.

MBAS_SecondPassComplete Data capture and correlation analysis is now complete.

Two data vectors, correlation amplitudes and path lengths, are ready for post processing.

MLS Sequence Generation

Random bit sequences can be generated by the recurrence relation:

where the generator polynomial M = 0x237 in the Micronas case.

The bit sequence is generated by taking the successive values of sn Sc I .

CLAIMS

1. A method for use in setting up a speaker system that comprises an array of output transducers capable of generating plural directed beams of sound, said method comprising: - emitting a test signal in the form of a directed sound beam from the speaker system into a room;

- detecting a reflection of said sound beam test signal at a location in said room; and

- determining a time of a maximum correlation between said emitted sound beam test signal and said reflected sound beam test signal.

2. A method according to claim 1, wherein said determined time of maximum correlation is used to calculate a path length for said directed sound beam.

3. A method according to claim 1 or 2, wherein said time of maximum correlation is obtained by convolving the reflected test signal detected at said location in the room with the emitted test signal and determining the time at which said maximum correlation occurs.

4. A method according to any one of claims 1 to 3, wherein said method is repeated for a plurality of angles and the magnitude of the maximum correlation between said emitted sound beam test signal and said reflected sound beam test signal is determined for each of said plurality of angles.

5. A method according to claim 4, wherein the test signal in the form of a directed sound beam is swept through a range of angles.

6. A method according to any one of claims 1 to 5, further comprising storing, for each emission angle of said directed sound beam test signal, the time of said maximum correlation and preferably also the magnitude of said maximum correlation.

7. A method according to any one of claims 4 to 6, wherein said angles are measured with respect to a line passing through the centre of said speaker system and being perpendicular to the front face of said speaker system.

8. A method according to any one of the preceding claims, further comprising initially determining the position of said location in said room relative to said speaker system by detecting direct, non-reflected, test signals at said location in said room.

9. A method according to any one of the preceding claims, wherein a sound beam direction for a centre channel is selected, which beam direction points from said speaker system directly towards said location in said room.

10. A method according to any one of the preceding claims, wherein a sound beam direction for a centre channel is selected to be the direction for which a maximum correlation exists between said emitted sound beam test signal and a directly received sound beam test signal at said location in said room.

11. A method according to any one of the preceding claims, wherein those angles for which a peak of maximum correlation exists are identified.

12. A method according to claim 11 , wherein a peak is defined as existing for an angle at which the maximum correlation is greater than the maximum correlation for either of the two neighbouring angles.

13. A method according to any one of the preceding claims, wherein said determined time of maximum correlation is used to select the angle of at least one of the surround sound channels.

14. A method according to any one of the preceding claims, wherein a sound beam direction for each of a left sound channel and a right sound channel is selected, which direction corresponds to an angle at which there is a peak magnitude of maximum correlation to the left and right of the centre channel, respectively.

15. A method according to any one of the preceding claims, wherein a sound beam direction for a left sound channel is selected, which direction corresponds to the angle at which there is the largest peak magnitude of maximum correlation out of all of the interrogated angles to the left of the centre channel.

16. A method according to any one of the preceding claims, wherein a sound beam direction for a right sound channel is selected, which direction corresponds to the angle at which there is the largest peak magnitude of maximum correlation out of all of the interrogated angles to the right of the centre channel.

17. A method according to any one of claims 14 to 16, wherein if the peak magnitude of maximum correlation for the left sound channel is smaller than a predetermined percentage of the peak magnitude of maximum correlation for the centre channel, then the speaker system is switched to "direct mode".

18. A method according to any one of claims 14 to 17, wherein if the peak magnitude of maximum correlation for the right sound channel is smaller than a predetermined percentage of the peak magnitude of maximum correlation for the centre channel, then the speaker system is switched to "direct mode".

19. A method according to any one of the preceding claims, wherein a sound beam direction for the left surround sound channel is selected, which direction corresponds to an angle at which there is a peak magnitude of maximum correlation to the left of the centre channel and which direction has a smaller beam angle and larger time of maximum correlation than the selected direction for the left sound channel.

20. A method according to any one of the preceding claims, wherein a sound beam direction for the right surround sound channel is selected, which direction corresponds to an angle at which there is a peak magnitude of maximum correlation to the right of the centre channel and which direction has a smaller beam angle and larger time of maximum correlation than the selected direction for the right sound channel.

21. A method according to any one of the preceding claims, further comprising determining a plurality of candidate beam directions for the surround sound channels and assigning a quality coefficient to each candidate.

22. A method according to claim 21, wherein the quality coefficient is determined by multiplying the maximum correlation value for the beam angle by the beam path length.

23. A method according to claim 22, wherein the path length used when calculating the quality coefficient is limited to some multiple of the path length calculated for the left sound channel or right sound channel, respectively.

24. A method according to any one of claims 21 to 23, wherein a candidate beam is disregarded, preferably by setting the quality coefficient to be zero, if any one of the following conditions (a) to (e) exist:

(a) the magnitude of maximum correlation for the candidate beam direction is less than a predetermined percentage of the magnitude of maximum correlation for the beam direction of the centre channel;

(b) the candidate beam has a larger angle and a larger time of maximum correlation than an already selected left or right sound channel;

(c) the candidate beam has a smaller angle and a smaller time of maximum correlation than an already selected left or right sound channel;

(d) the candidate beam and an already determined candidate beam have a time of maximum correlation within a predetermined amount of one another and the candidate beam and said already determined candidate beam have an emission angle within a predetermined amount of one another;

(e) the candidate beam is directed in a direction where the path length for adjacent angles differs by a predetermined amount from the candidate beam path length.

25. A method according to any one of claims 21 to 24, wherein the candidate beam direction having the largest quality coefficient is selected for a surround sound channel.

26. A method according to any one of the preceding claims, wherein the focal length for the centre channel is set to be a fixed negative number.

27. A method according to any one of the preceding claims, wherein the focal length for the left sound channel and/or the right sound channel is set based on the path lengths for those channels, preferably being set at one half of the path length.

28. A method according to any one of the preceding claims, wherein a correlation is calculated only for those sound beams having a path length larger than the distance from the speaker system to the location in said room.

29. A method according to any one of the preceding claims, whereby, if the angle between the identified left sound channel and identified right sound channel is smaller than a predetermined angle, a "corner mode" is set for the speaker system.

30. A method according to claim 29, wherein when a "corner mode" is set, the left sound channel and right sound channel are emitted in the "direct mode".

31. A method according to claim 17, 18 or 30, wherein, in the "direct mode", the left sound channel and right sound channel are directed in the same direction as the centre sound channel.

32. A method to claim 17, 18, 30 or 31, wherein, in the "direct mode", a subset of the transducers in the array are used to emit the left sound channel and right sound channel.

33. A method according to any one of the preceding claims, wherein the test signal comprises a maximum length sequence.

34. A method according to any one of the preceding claims, wherein the correlation is calculated in real time, in the time between samples of the test signal.

35. A method according to any one of the preceding claims, wherein said reflection of said sound beam test signal is detected at the listening position by an omnidirectional microphone.

36. A method of outputting surround sound, said method comprising: performing a method comprising the method of any one of the preceding claims; determining beam emission angles for a centre, left, right, surround left and surround right channel respectively; emitting sound beams representing said respective channels in directions given by said respective angles.

37. Apparatus for outputting surround sound that is constructed and arranged to perform the method of any one of the preceding claims.

38. Apparatus for outputting sound, said apparatus comprising: an array of output transducers; an input transducer; and a controller programmed: to cause said array to emit a test signal in the form of a directed sound beam into a room; to receive a signal from said input transducer that corresponds to at least one reflection of said sound beam test signal; and to determine a time of a maximum correlation between said emitted sound beam test signal and said received reflected sound beam test signal.

39. An apparatus according to claim 38, wherein said input transducer is an omnidirectional microphone for placement at the listening position.

40. An apparatus according to claim 38 or 39, wherein said controller is arranged to calculate a path length for said directed sound beam.

41. An apparatus according to any one of claims 38 to 40, wherein said controller is arranged to sweep said directed sound beam through a range of angles and to determine the magnitude of the maximum correlation between said emitted sound beam test signal and said reflected sound beam test signal for each angle in said range.

42. An apparatus according to any one of claims 38 to 41, further comprising a memory for storing, for each emission angle of said directed sound beam test signal, the time of said maximum correlation and preferably also the magnitude of said maximum correlation.

43. An apparatus according to any one of claims 38 to 42, wherein said controller is arranged to initially determine the position of said location in said room relative to said speaker system by receiving signals from said input transducer relating to direct, non- reflected, test signals.

44. An apparatus according to any one of claims 38 to 43, wherein said controller is arranged to select a sound beam direction for a centre channel, which beam direction points from said speaker system directly towards said location in said room.

45. An apparatus according to any one of claims 38 to 44, wherein said controller is arranged to select a sound beam direction for a centre channel, which beam direction is set to be the direction for which a maximum correlation exists between said emitted sound beam test signal and a directly received sound beam test signal by said input transducer.

46. An apparatus according to any one of claims 38 to 45, wherein said controller is arranged to select the angle for at least one of the surround sound channels based on the determined time of maximum correlation.

47. An apparatus according to any one of claims 38 to 46, wherein the controller is arranged to identify those angles for which a peak of maximum correlation exists.

48. An apparatus according to any one of claims 38 to 47, wherein said controller is arranged to select a sound beam direction for a left sound channel, which direction corresponds to the angle at which there is the largest peak magnitude of maximum correlation out of all of the interrogated angles to the left of the centre channel.

49. An apparatus according to any one of claims 38 to 48, wherein said controller is arranged to select a sound beam direction for a right sound channel, which direction corresponds to the angle at which there is the largest peak magnitude of maximum correlation out of all of the interrogated angles to the right of the centre channel.

50. An apparatus according to any one of claims 38 to 49, wherein said controller is arranged to switch the speaker system to "direct mode" if the peak magnitude of maximum correlation for the left sound channel is smaller than a predetermined percentage of the peak magnitude of maximum correlation for the centre channel.

51. An apparatus according to any one of claims 38 to 50, wherein said controller is arranged to switch the speaker system to "direct mode" if the peak magnitude of maximum correlation for the right sound channel is smaller than a predetermined percentage of the peak magnitude of maximum correlation for the centre channel.

52. An apparatus according to any one of claims 38 to 51, wherein said controller is arranged to select a sound beam direction for the left surround sound channel, which direction corresponds to an angle at which there is a peak magnitude of maximum correlation to the left of the centre channel and which direction has a smaller beam angle and larger time of maximum correlation than the selected direction for the left sound channel.

53. An apparatus according to any one of claims 38 to 52, wherein said controller is arranged to select a sound beam direction for the right surround sound channel, which direction corresponds to an angle at which there is a peak magnitude of maximum correlation to the right of the centre channel and which direction has a smaller beam angle and larger time of maximum correlation than the selected direction for the right sound channel.

54. An apparatus according to any one of claims 38 to 53, further comprising a candidate beam determining means for determining a plurality of candidate beam directions for the surround sound channels and assigning a quality coefficient to each candidate.

55. An apparatus according to claim 54, wherein the quality coefficient is determined by multiplying the maximum correlation value for the beam angle by the beam path length.

56. An apparatus according to claim 54 or 55, wherein said candidate beam determining means is arranged to disregard a candidate beam, preferably by setting the quality coefficient to be zero for that beam, if any one of the following conditions (a) to (e) exist:

(a) the magnitude of maximum correlation for the candidate beam direction is less than a predetermined percentage of the magnitude of maximum correlation for the beam direction of the centre channel;

(b) the candidate beam has a larger angle and a larger time of maximum correlation than an already selected left or right sound channel;

(c) the candidate beam has a smaller angle and a smaller time of maximum correlation than an already selected left or right sound channel;

(d) the candidate beam and an already determined candidate beam have a time of maximum correlation within a predetermined amount of one another and the candidate beam and said already determined candidate beam have an emission angle within a predetermined amount of one another; (e) the candidate beam is directed in a direction where the path length for adjacent angles differs by a predetermined amount from the candidate beam path length.

57. An apparatus according to any one of claims 54 to 56, wherein the controller is arranged to select the candidate beam direction having the largest quality coefficient for a surround sound channel.

58. An apparatus according to any one of claims 38 to 57, wherein the controller is arranged to set a "corner mode" for the speaker system if the angle between the identified left sound channel and identified right sound channel is smaller than a predetermined angle.

59. An apparatus according to claim 58, wherein the controller is arranged to emit the left sound channel and right sound channel in "direct mode" if the "corner mode" has been set for the speaker system.

60. An apparatus according to claim 50, 51 or 59, wherein, in the "direct mode", the controller is arranged to direct the left sound channel and right sound channel in the same direction as the centre sound channel.

61. An apparatus according to claim 50, 51, 59 or 60, wherein the speaker system is arranged to use a subset of the transducers in the array to emit the left sound channel and right sound channel when in the "direct mode".

62. An apparatus according to any one of claims 38 to 61, wherein the controller is arranged to calculate the correlation in real time, in the time between samples of the test signal.

Sign in to the Lens