WO2007125449A1

WO2007125449A1 - Device for and method of estimating directivity of a microphone arrangement

Info

Publication number: WO2007125449A1
Application number: PCT/IB2007/051324
Authority: WO
Inventors: Daniel W. E. Schobben
Original assignee: Koninklijke Philips Electronics N.V.
Priority date: 2006-04-28
Filing date: 2007-04-12
Publication date: 2007-11-08

Abstract

A device (120) for quantifying directivity of a microphone arrangement (110), the device (120) comprising a first input unit (121) for inputting a useful signal (S) generated by the microphone arrangement (110) in the presence of a useful sound signal source (102) and in the absence of a noise source (201), a second input unit (122) for inputting a noise signal (N) generated by the microphone arrangement (110) in the absence of the useful sound signal source (102) and in the presence of the noise source (201), a determining unit (304) adapted to determine a signal-to-noise ratio (SNR) of the microphone arrangement (110) on the basis of the useful signal (S) and on the basis of the noise signal (N), and an output unit (305) for outputting the determined signal-to-noise ratio (SNR) of the microphone arrangement (110) as a quantitative measure of directivity of the microphone arrangement (110).

Description

Device for and method of estimating directivity of a microphone arrangement

FIELD OF THE INVENTION

The invention relates to a device for estimating directivity of a microphone arrangement.

The invention also relates to a method of estimating directivity of a microphone arrangement.

Furthermore, the invention relates to a program element and to a computer- readable medium.

The invention further relates to a method of use.

BACKGROUND OF THE INVENTION

Audio playback devices are becoming more and more important. Particularly, an increasing number of users buy microphone arrangements having a directivity feature for proper audio quality, even in noisy environments.

Directivity of audio signal capturing may be a feature of a microphone or microphone array.

WO 2005/029914 discloses a signal-processing apparatus for a hearing aid with a controllable directional characteristic which comprises a directional controller receiving first and second microphone signals and outputs an output signal, a signal analyzer which detects whether at least one of the first and second microphone signals is an undesired signal, and wherein said directional controller minimizes the output signal by adjusting the directional characteristic only if the signal analyzer has detected undesired signals. An index referred to as Directivity Index (DI) allows quantifying the directivity of a microphone arrangement. It is conventionally defined as the logarithmic ratio of the power supplied from an omni-directional microphone arrangement to the power supplied from a directional microphone arrangement with equal sensitivity in the principal direction, in a diffuse noise field.

However, this classification may be an inappropriate measure of directivity of an adaptive microphone arrangement. OBJECT AND SUMMARY OF THE INVENTION

It is an object of the invention to provide a reliable measure of directivity of a microphone arrangement.

In order to achieve the object defined above, a device for estimating directivity of a microphone arrangement, a method of estimating directivity of a microphone arrangement, a program element, a computer-readable medium, and a method of use as defined in the independent claims are provided.

In accordance with an embodiment of the invention, a device for estimating (or determining or quantifying) directivity of a microphone arrangement is provided, the device comprising a first input unit for inputting a useful signal generated by the microphone arrangement in the presence of a useful sound signal source and in the absence of a noise source, a second input unit for inputting a noise signal generated by the microphone arrangement in the absence of the useful sound signal source and in the presence of the noise source, a determining unit adapted to determine a signal-to-noise ratio of the microphone arrangement on the basis of the useful signal and on the basis of the noise signal after processing them individually by the shadow filters, and an output unit for outputting the determined signal-to-noise ratio of the microphone arrangement as a (qualitative or quantitative) measure of directivity of the microphone arrangement.

In accordance with another embodiment of the invention, a method of estimating directivity of a microphone arrangement is provided, the method comprising the steps of generating a useful signal by the microphone arrangement in the presence of a useful sound signal source and in the absence of a noise source, generating a noise signal by the microphone arrangement in the absence of the useful sound signal source and in the presence of the noise source, determining a signal-to-noise ratio of the microphone arrangement on the basis of the useful signal and on the basis of the noise signal, and outputting the determined signal-to-noise ratio of the microphone arrangement as a (qualitative or quantitative) measure of directivity of the microphone arrangement.

In accordance with a further embodiment of the invention, a signal-to-noise ratio estimated for a microphone arrangement is used (or interpreted or defined) as a measure of directivity of the microphone arrangement.

In accordance with a further embodiment of the invention, a program element is provided, which, when being executed by a processor, is adapted to control or carry out a method of estimating directivity of a microphone arrangement having the above-mentioned features. In accordance with yet another embodiment of the invention, a computer- readable medium is provided, in which a computer program is stored which, when being executed by a processor, is adapted to control or carry out a method of estimating directivity of a microphone arrangement having the above-mentioned features. The data-processing in accordance with embodiments of the invention can be realized by a computer program, i.e. by software, or by using one or more special electronic optimization circuits, i.e. in hardware, or in a hybrid form, i.e. by means of software and hardware components.

The term "directivity" may particularly denote the property of a microphone or microphone array including analog or digital signal-processing of being more sensitive in one direction than in another.

The term "useful sound source" may particularly denote a sound source which is desired in a special application to be the only and undisturbed sound source. This may be a person giving a speech, a person talking on a hands-free mobile phone while simultaneously driving a car, or a piano the sound of which is to be recorded. In an ideal situation, such a useful, desired or wanted signal is not disturbed by any noise or other interfering sound components at all.

The term "useful signal" or "wanted signal" may particularly denote a signal being the contrast of a disturbing signal. In communication techniques, it is usually desired to have a useful signal that is as strong as possible, whereas it is usually desired to keep the disturbing signal as small as possible. A useful signal usually carries information, whereas a disturbing signal or a noise signal is usually free of any information and structure such as road, wind, or background noise. The disturbing signal may also be, for example, a concurring speaker. In the context of this application, the term "noise signal" may particularly denote any disturbing signal, particularly any signal without a specific meaning, content or information, wherein such a noise signal may have isotropic and statistic properties.

In the context of this application, the term "microphone" may particularly denote a device that records sound and converts sound into electric signals. This may or may not include further processing, which may be adaptive processing.

In the context of this application, the term "microphone array" may particularly denote a set of devices each recording sound and converting sound into electric signals.

In the context of this application, the term "shadow filter" may particularly denote an individual filter that has (always) the same filter coefficients as a primary filter (for example, shares filter coefficients with an adaptive filter). It may be a new example of a filter that shares filter coefficients with another filter.

In the context of this application, the term "adaptive" may particularly denote a filter that has a time- varying behavior, and may adapt to signals. For beam- forming, adaptive filter may try to weight sound from different directions (for example, concentrate on a desired direction, suppress undesired directions, etc.).

In accordance with an embodiment of the invention, a meaningful classification or quantification of the degree of directivity of a microphone array including one or more microphones and the corresponding signal-processing system may be provided. When referred to as "directivity of a microphone array", this may relate to the directivity of the microphone array including the corresponding signal-processing system. "Signals recorded from the microphone array" may be understood to be the individual unprocessed signals of all the individual microphones in the array. In contrast to conventional approaches, the signal-to-noise ratio (when the microphone (array) captures a pure wanted signal, for example, a speech signal, as compared to a situation in which the same microphone (array) captures a pure noise signal) is determined. This classification may allow a more reasonable and technically correct definition of the directivity of a microphone array. On the basis of such a classification or quantification scheme, a proper basis may therefore be provided to allow a technician to select a particular microphone array for a special technical application, based on such a meaningful directivity index or value.

In accordance with an embodiment of the invention, the directivity index may be redefined particularly for adaptive microphone arrays. A method is provided for defining a directivity index of an adaptive microphone array by determining a signal-to-noise ratio of the array, for example, by measuring a ratio between a speech signal recorded by the array in the absence of other sound sources, and a noise signal recorded by the array in the absence of the speech signal. Prior to computing this ratio, the individual signals are processed by shadow filters of the DUT that works on the sum of the individual signals.

The traditional definition of the Directivity Index (DI) described above allows quantifying the directivity of a microphone or a non-adaptive array of microphones assuming diffuse noise conditions. For a two -microphone array, the theoretical upper boundary of the directivity index may equal 6dB. For adaptive microphone arrays, the DI may not reflect the real-time performance of the array with theoretical DI values that can be driven to infinity for spatially separated sound sources in anechoic conditions. In product specifications, DIs have been reported for an adaptive two -microphone array up to 17dB.

In accordance with an embodiment of the invention, the DI may be redefined particularly for adaptive arrays as an objective measure to quantify improvements in signal- to-noise ratio based on directivity. In accordance with an embodiment, a device for determining a signal-to-noise ratio (SNR) of (for example, adaptive) microphone arrays is provided. Such a device may have a first input adapted to receive a speech signal (more generally a useful signal) and a second input adapted to receive a noise signal. The speech signal may be representative of a desired sound source obtained in isolation for all microphones in a microphone array in the absence of other sound sources. The noise signal may be representative of background noise obtained in isolation for all microphones in the microphone array in the absence of the desired sound source.

In accordance with an embodiment, such a device may further comprise an adaptive filter, a first shadow filter and a second shadow filter. A calculation unit may be provided and adapted to supply an output signal which is indicative of the signal-to-noise ratio, wherein the adaptive filter may be adapted to receive a summed signal of the speech signal and the noise signal, while the first and the second shadow filter may be adapted to receive filter coefficients of the adaptive filter. The first shadow filter may be adapted to receive the speech signal for generating a first filtered signal. The second shadow filter may be adapted to receive the noise signal for generating a second filtered signal. The calculation unit may be adapted to supply the output signal on the basis of the ratio of the first and the second filtered signal.

In accordance with a further embodiment, the signal-to-noise ratio can again be averaged over time to obtain an average signal-to-noise ratio. The average time may be a fixed time stored in the device, or may be determined to have a reasonable duration to obtain a meaningful averaging.

In accordance with an embodiment, the signal-to-noise ratio may be better assessed as a function of time when predicting speech intelligibility in a situation with, for example, concurring speakers. Microphone arrays are beneficial in applications that require a high signal-to- noise ratio (SNR), such as hands-free telephony, continuous speech recognition and hearing aids. Further SNR improvements may be obtained by adaptively filtering the microphone signals, for example, preserving the desired sound source while suppressing interfering sound sources and background noise. A reasonably defined Directivity Index (DI) should allow quantifying the directivity of a microphone or a microphone array.

The ANSI S3.35-2004 standards ("Method of Measurement of Performance Characteristics of Hearing Aids Under Simulated Real-Ear Working Conditions") provide procedures for obtaining the directional response of an acoustical manikin as a function of azimuth and elevation of the sound source, with and without the assistance of a hearing aid, and for calculating the Directivity Index from the directional response. In the absence of a standard for assessing adaptive microphone arrays, this standard has been used for time- varying systems as well. The results may not always adequately reflect the potential of adaptive microphone arrays. The assumptions made in the above conventional definition of the DI include weaknesses, particularly when being applied to estimating SNR improvements obtained from a directional microphone (array). These assumptions include:

- the desired sound source enters the array through the beam as a principal direction; - the noise can be represented as a diffuse noise field;

- both noise and desired sound are stationary signals;

- the directivity of the microphone array is fixed.

Embodiments of the present invention are based on the recognition that these conventional assumptions have, inter alia, the drawback that the desired sound source may enter the array from any direction through reflections or because the apparent sound source is relatively large. The traditional definition of DI favors a narrow beam width, while in practice a somewhat wider beam width may capture more of the desired sound source. A modified Directivity Index calculation that partly solves this issue is introduced in J. Beard, "Modified Directivity Index Calculation", Internal Communications, Knowles Electronics Inc., April 6, 1999 and at http://www.knowlesacoustics.com/images/pdf/white/AdvancedMicTechnology.pdf. The conventional definition has the further drawback that the diffuse noise field assumption only holds for very specific cases. In practical cases, which include a distracter which is spatially apart from the desired sound source, the specific array response with its side lobes and nulls has a great impact on the SNR improvements obtained.

A further drawback is that the SNR may be better assessed as a function of time when predicting speech intelligibility in situations with, for example, concurring speakers.

Moreover, the strength of adaptive arrays may particularly reside in suppressing a distracter for which theoretically perfect suppression can be achieved which is not reflected in the traditional DI.

Based on the recognition of these and other drawbacks, embodiments of the invention suggest a redefinition of the DI, allowing a technically more reasonable classification of microphone arrays and specification of microphone products so as to simplify selection of a microphone array product for a specific application based on the definition of directivity in accordance with embodiments of the invention.

In accordance with an embodiment, a Directivity Index definition for time- varying systems may be provided. This DI definition and determination scheme is based on the fact that sound is additive so that the desired sound source can be recorded in isolation for all microphones in the array in the absence of other sound sources (for example, in an empty room, a church, or a cafeteria). In a later (or previous) phase, background noise may be recorded in isolation (for example, at a cocktail party in the same room, in the crowded church, or in the busy cafeteria). The sound to which the microphone array must adapt is the sum of the desired sound S and the background noise recording N. S and N may be scalar values (for a single microphone apparatus) or vectors (for a multiple microphone apparatus) containing the signals of all microphones, in a scenario in which a plurality of microphones is present.

The method may be based on a similar technique as may be used to assess the performance of blind signal separation methods (see D.W.E Schobben, K. Torkkola, P.

Smaragdis, "Evaluation of Blind Signal Separation Methods", in Proceedings Int. Workshop Independent Component Analysis and Blind Signal Separation, Aussois, France, January 11- 15, 1999, pages 261-266).

In accordance with an embodiment of the invention, the SNR of the adaptive microphone array is determined by running two shadow filters in parallel with the adaptive filter. The shadow filters have the same filter coefficients as the adaptive filters at all times. The adaptive filter may be fed with the signal S+N, whereas the shadow filters are fed with S and N, respectively. The ratio of the power of the outputs of the shadow filters may now provide the true SNR. If desired, the SNR can again be averaged over time to obtain an average SNR. Such a method may overcome the drawbacks associated with the assumptions that are made in the traditional DI definition.

Thus, the redefined Directivity Index (particularly) for time- varying systems (DI-TV) can be used to assess adaptive microphone arrays in virtually any real-life situation. For fixed microphone arrays, the new definition defaults to a further improvement of the modified Directivity Index calculation by treating all sound that is emitted by the desired sound source as signal, while treating all background noise as noise.

In accordance with an embodiment, this definition may be used in the context of a standard (for example, ANSI, ITU) defining the directivity of devices with microphone arrays in an embodiment of the invention.

In accordance with an embodiment, a measurement procedure for estimating the DI value may include that the target sound sources are recorded in isolation for all microphones. The background noise may be recorded in isolation for all microphones. The adaptive system may work on the summed microphone signals for target source and noise. The filter coefficients in the adaptive systems may be copied into two shadow filters which process both target and noise independently. The resulting signals of the previous step may be used for assessing the performance (for example, directivity) of the adaptive filter, for example, by evaluating the relative power.

Fields of application of embodiments of the invention are standards for assessing the directivity and performance of mobile phones, headsets, hearing aids, and hands- free systems.

Further embodiments of the device according to the invention for estimating/quantifying directivity of a microphone arrangement will be explained hereinafter. However, these embodiments also apply to the method of estimating/quantifying directivity of a microphone arrangement, the program element and the computer-readable medium.

Particularly the useful signal may be a speech signal, for example, of a person talking, or the like. Alternatively, the useful signal may be a music signal, or any other audio signal to be recorded.

The first input unit of the device may comprise a plurality of first input channels each adapted to input a useful signal generated by a respective one of a plurality of microphones of the microphone arrangement in the presence of the useful sound signal source and in the absence of the noise source. When a multi-microphone arrangement is analyzed or categorized, the recorded input of each microphone may thus be input separately into the device, for example, for separate processing. This may allow defining a Directivity Index for the entire microphone arrangement with a high accuracy and resolution.

Also the second input unit may comprise a plurality of second input channels each adapted to input a noise signal generated by a respective one of the plurality of microphones of the microphone arrangement in the absence of the useful sound signal source and in the presence of the noise source. Also the noise signal may therefore be captured separately for each microphone and may be input, separately for the different microphones, into the device.

The determining unit may be adapted to determine the signal-to-noise ratio of the microphone arrangement averaged over a time duration of the useful signal and the noise signal. Depending on the time dependence of particularly the useful signal (and perhaps also of the noise signal), the estimated value for indicating the directivity may vary (slightly) in time. Therefore, a more meaningful result may be obtained when such instantaneous directivity values are averaged over a sufficiently long period of time. This may further increase the accuracy of the derived directivity value. The device may comprise a first filter unit adapted to filter the useful signal under the control of a control signal, and a second filter unit adapted to filter the noise signal under the control of the control signal. A third filter unit may be provided for filtering a sum of the useful signal and the noise signal and may thereby generate the control signal. In other words, the third filter unit may be coupled to an output of an adding unit which adds the noise signal and the useful signal. Based on this sum signal, the third filter unit may derive filter parameters for the first and the second filter unit, which parameters may be encoded in the control signal. This control signal may be fed into the first and the second filter unit, allowing a preferably synchronized filtering of the useful signal and the noise signal by the first and the second filter unit. The first filter unit and/or the second filter unit may be shadow filters. The third filter unit may be an adaptive filter.

The first and the second filter unit may be coupled to the determining unit so that signals supplied at outputs of the first and the second filter unit are supplyable to the determining unit for determining the signal-to-noise ratio. In other words, the determining unit may receive a filtered useful signal and a filtered noise signal and may derive the directivity from the power of these signals.

The determining unit may be adapted to determine the signal-to-noise ratio as a ratio between the signals supplied at the outputs of the first and the second filter unit. This ratio may be formed as a ratio of the amplitudes, intensities or powers between the supplied signals.

Advantageously, the filter parameters of the first, the second and the third filter unit may be identical. Such a synchronized operation of the filter units may allow obtaining directly comparable signals.

The output unit may be adapted to output a directivity index as the quantitative measure of the directivity of the microphone arrangement. Thus, at the output of the device, a simple value, for example, a number may be provided which is a reliable measure of the directivity characteristics of the microphone arrangement.

The device may be adapted to quantify directivity of a microphone arrangement comprising a plurality of microphones. Arrangements with multi-microphones may have a particular directivity feature. However, embodiments of the invention may also be implemented in the context of single microphone arrangements.

The device may be adapted to quantify directivity of an adaptive microphone array. However, also simple microphone arrays without adaptive capabilities may be classified in accordance with the functionality of the device.

The apparatus may be adapted to quantify directivity of a mobile phone, a headset, a loudspeaker, a hearing aid, a handsfree system, a television device, a video recorder, a monitor, a gaming device, a laptop, an audio player, a DVD player, a CD player, a hard disk-based media player, an Internet radio device, a public entertainment device, an MP3 player, a hi-fi system, a vehicle entertainment device, a car entertainment device, a medical communication system, a body-worn device, a speech communication device, a home cinema system, or a music hall system. A "car entertainment device" may be a hi-fi system for an automobile.

However, although the system according to the invention primarily intends to quantify directivity of audio recording systems, it is also possible to apply the system for a combination of audio and video data. For example, an embodiment of the invention may be implemented to classify audiovisual application devices such as a video camera in which a microphone array is used.

These and other aspects of the invention are apparent from and will be elucidated with reference to the embodiments described hereinafter.

BRIEF DESCRIPTION OF THE DRAWINGS

In the drawings,

Fig. 1 shows a first experimental environment for capturing pure speech audio signals by a microphone arrangement for quantifying directivity of the microphone arrangement in accordance with an embodiment of the invention.

Fig. 2 shows a second experimental environment for capturing pure noise audio signals by the microphone arrangement of Fig. 1 for quantifying directivity of the microphone arrangement in accordance with an embodiment of the invention. Fig. 3 shows an embodiment of a device according to the invention for quantifying directivity of the microphone arrangement of Figs. 1 and 2.

DESCRIPTION OF EMBODIMENTS The illustrations in the drawings are schematic. In the different Figures, similar or identical elements are denoted by the same reference signs.

An embodiment of a device 120 according to the invention for quantifying directivity of a microphone arrangement 110 will now be described with reference to Figs. 1 to 3. Before describing the function of the device 120 in detail, a first experimental setup 100 and a second experimental setup 200 will be described with reference to Figs. 1 and 2 in which signals are captured which are then used by the device 120.

Fig. 1 shows an experimental setup 100 in which the microphone array 110 is positioned within an acoustically isolated room 101. A target sound source 102 (for example, a human speaker talking to emit acoustic waves 103) is positioned at a particular location within the room 101 to emit useful sound waves 103. These useful sound waves 103 have some kind of directivity, i.e. they are distributed in a non-isotropic manner in the room 101, and captured by the individual microphones 111, 112, 113, ..., 114 of the microphone arrangement 110. A recording device 115, for immediate or later processing, records the useful sound signals captured by each microphone 111 to 114. These useful signals may be supplied to a first input unit 121 of the device 120.

As is shown in Fig. 2, in the second experimental setup 200, a noisy sound source 201 (for example, a plurality of human speakers talking in an unspecifϊc manner to emit acoustic waves 202) emits noise acoustic waves 202, typically in an isotropic manner in the room 101. In other words, in this scenario, the background noise is recorded in isolation for all microphones 111 to 114, and the respective noise channel signals are supplied to the recording device 115 for immediate or later processing. The recording device 115 is or may be coupled to a second input unit 122 of the device 120 so as to supply the signals received at the outputs of the microphones 111 to 114.

A detailed view of the device 120 for quantifying directivity of the microphone arrangement 110 will now be explained in more detail with reference to Fig. 3.

The device 120 comprises the first input unit 121 for inputting the useful signal generated by the microphone arrangement 110 in the presence of the useful sound signal source 102 and in the absence of the noise source 201 (see scenario of Fig. 1).

Furthermore, the device 120 comprises the second input unit 122 for inputting the noise signal generated by the microphone arrangement 110 in the absence of the useful sound signal source 102 and in the presence of the noise source 201 (see scenario of Fig. 2). The useful signal S is thus supplied via the first input unit 121, and the useful signal N is supplied via the second input unit 122 to the device 120.

The signal flows of Fig. 3 contain signals of all microphones 111 to 114. Filters 301 to 303, which will be explained hereinafter, may be multi-channel filters.

The device 120 includes a summing unit 300 for summing the useful signal S and the noise signal N so that a summed signal S+N may be generated and input to an adaptive filter unit 302. The useful signal S is supplied to a first shadow filter unit 301. The noise signal N is supplied to a second shadow filter unit 303. The adaptive filter unit 302 generates a control signal 304 which is supplied to both shadow filter units 301, 303 for filtering the respective useful signal S and noise signal N in a synchronized manner. The system 120 thus works on the summed microphone signals S and N for the target source 102 and for the noise source 201.

Filter coefficients in the adaptive system 120 are copied into the two shadow filter units 301 and 303 which process both target signal S and noise signal N simultaneously and independently. Processed signals s and n are provided at the output of the shadow filter units 301 and 303, respectively, which signals are supplied to a determining unit 304 for determining a signal-to-noise ratio SNR. In other words, the determining unit 304 is adapted to determine the signal-to-noise ratio SNR of the microphone arrangement 110 on the basis of the useful signal S and the noise signal N. At an output 305, the determined signal-to-noise ratio SNR of the microphone arrangement 110 is provided as a quantitative measure of directivity of the microphone arrangement 110.

The sum signal S+N is provided at a further output (indicated in Fig. 3 by an arrow connected to the adaptive filter 302 and pointing to the right-hand side) of the adaptive filter 302. This signal S+N is not necessary for determining the Directivity Index, but is a signal which a user of a microphone array 110 is usually interested in (for example, a signal which may be sent to a loudspeaker for audible reproduction of the audio content).

Although not shown in Fig. 3, a directivity value calculation unit may be provided between the determining unit 304 and the output 305 for determining a parameter value for the directivity of the microphone arrangement 110 that is directly derivable from the signal-to-noise ratio SNR. Optionally, the SNR value may be averaged over a certain time interval so as to derive more meaningful information suppressing time-dependent artefacts. The filter parameters for the filter units 301 to 303 may be the same.

Furthermore, E {....} (for example, E{s^Λ2/n^Λ2}) may be assigned to the determining unit 304 and may denote an expectation operator, which may be computed as a running average. The term "running average" denotes that an averaged value may be calculated continuously or several times. In other words, it may denote a time-dependent average value calculated or updated again and again. A running average function may keep a running total of all readings, and may divide this by the number of readings captured. A method of redefining the directivity index for adaptive microphone arrays in accordance with an embodiment of the invention will now be described in further detail.

The directivity index (DI) allows quantifying the directivity of a microphone or a non-adaptive array of microphones, which indicates how well a target sound source can be extracted in the presence of diffuse background noise. Adaptive microphone arrays have the potential to provide an improved performance for suppressing distracters that are spatially apart from the target sound source. In the absence of performance measures for adaptive microphone arrays, the DI may be used to evaluate adaptive microphone arrays in both anechoic and diffuse noise conditions. However, for adaptive microphone arrays, the DI may not reflect the real- life performance of the array with theoretical DI values that can be driven to infinity for spatially separated target and distracter sound sources in anechoic conditions, while such conditions will hardly be encountered in daily life. On the other hand, adaptive microphone arrays inherently do not have to improve performance in diffuse sound fields for which a DI has originally been defined. Therefore, in accordance with an embodiment, the DI may be redefined for adaptive arrays as an objective measure to quantify improvements in signal-to-noise based on directivity.

In microphone array applications requiring a high signal-to-noise ratio, "signal" may relate to a desired speech signal and "noise" may relate to all other sounds including background noise and undesired concurring speakers, for example, in a cocktail party scenario. The SNR can be improved or optimized by adaptively filtering the microphone signals so as to preserve the desired sound source while suppressing interfering sound sources and background noise. The directivity that is obtained from an adaptive microphone array may be assessed and as such may provide a valuable tool for assessing and comparing performance improvements of adaptive microphone arrays.

The ultimate performance measure and continuous speech recognition relates to the speech recognition rate. In hearing aids, this may be speech intelligibility and comfort for the user, whereas in communication headsets performance may relate to speech intelligibility and comfort as perceived by the far-end speaker.

In accordance with an embodiment, the Directivity Index for time- varying systems may be provided in a manner as has been described above with reference to Figs. 1 to 3.

Fig. 3 shows a system for evaluating SNR improvement using prestored multi- microphone recordings of the desired speech S and noise N recorded in isolation.

The expectation operator E { ... } is computed in practice as a running average. The SNR computation may be replaced in this scheme by other performance evaluating schemes, some of which may require a close talking microphone to record a dry speech reference.

A Directivity Index for Time Varying systems (DI-TV) may thus be calculated to characterize directivity of a microphone array. The DI-TV can be used to assess adaptive microphone arrays in many real- life situations. For fixed microphone arrays, this definition defaults to a further improvement of the modified directivity index calculation by treating all sound that is emitted by the desired sound source as a signal while treating all background noise as noise rather than relying on sound within a certain angle around the principal access to be a signal and signals outside this angle to be noise. It should be noted that use of the verb "comprise" and its conjugations does not exclude other elements, steps or features and that use of the indefinite article "a" or "an" does not exclude a plurality. Also elements described in association with different embodiments may be combined.

It should also be noted that reference signs in the claims shall not be construed as limiting the scope of the claims.

Claims

CLAIMS:

1. A device (120) for estimating directivity of a microphone arrangement (110), the device (120) comprising a first input unit (121) for inputting a useful signal (S) generated by the microphone arrangement (110) in the presence of a useful sound signal source (102) and in the absence of a noise source (201); a second input unit (122) for inputting a noise signal (N) generated by the microphone arrangement (110) in the absence of the useful sound signal source (102) and in the presence of the noise source (201); a determining unit (304) adapted to determine a signal-to-noise ratio (SNR) of the microphone arrangement (110) on the basis of the useful signal (S) and on the basis of the noise signal (N); an output unit (305) for outputting the determined signal-to-noise ratio (SNR) of the microphone arrangement (110) as a measure of directivity of the microphone arrangement (110).

2. The device (120) according to claim 1, wherein the useful signal (S) is a speech signal.

3. The device (120) according to claim 1, wherein the first input unit (121) comprises a plurality of first input channels each adapted to input a useful signal (S) generated by a respective one of a plurality of microphones (111 to 114) of the microphone arrangement (110) in the presence of a useful sound signal source (102) and in the absence of a noise source (201).

4. The device (120) according to claim 1, wherein the second input unit (122) comprises a plurality of second input channels each adapted to input a noise signal (N) generated by a respective one of a plurality of microphones (111 to 114) of the microphone arrangement (110) in the absence of the useful sound signal source (102) and in the presence of the noise source (201).

5. The device (120) according to claim 1, wherein the determining unit (304) is adapted to determine the signal-to-noise ratio (SNR) of the microphone arrangement (110) averaged over a predetermined time duration of the useful signal (S) and the noise signal (N).

6. The device (120) according to claim 1, comprising a first filter unit (301) adapted to filter the useful signal (S) under the control of a control signal (304); a second filter unit (303) adapted to filter the noise signal (N) under the control of the control signal (304); a third filter unit (302) adapted to filter a sum of the useful signal (S) and of the noise signal (N), thereby generating the control signal (304).

7. The device (120) according to claim 6, wherein at least one of the group consisting of the first filter unit (301) and the second filter unit (303) is a shadow filter.

8. The device (120) according to claim 6, wherein the third filter unit (302) is an adaptive filter.

9. The device (120) according to claim 6, wherein the first filter unit (301) and the second filter unit (303) are coupled to the determining unit (304) so that signals (s, n) provided at outputs of the first filter unit (301) and the second filter unit (303) are supplyable to the determining unit (304) for determining the signal-to-noise ratio (SNR).

10. The device (120) according to claim 9, wherein the determining unit (304) is adapted to determine the signal-to-noise ratio (SNR) as a ratio between the signals (s, n) provided at the outputs of the first filter unit (301) and the second filter unit (303).

11. The device (120) according to claim 6, wherein filter parameters of the first filter unit (301), the second filter unit (303), and the third filter unit (302) are identical.

12. The device (120) according to claim 1, wherein the output unit (305) is adapted to output a directivity index as the measure of directivity o f the microphone arrangement (110).

13. The device (120) according to claim 1, adapted to estimate directivity of a microphone arrangement (110) comprising a plurality of microphones (111 to 114).

14. The device (120) according to claim 1, adapted to estimate directivity of an adaptive microphone array (110).

15. The device (120) according to claim 1, adapted to quantify the directivity of the microphone arrangement (110) by adapting the output unit (305) for outputting the determined signal-to-noise ratio (SNR) of the microphone arrangement (110) as a quantitative measure of directivity of the microphone arrangement (110).

16. The device (120) according to claim 1, adapted to estimate directivity of a microphone arrangement (110) comprising apparatuses of the group consisting of a mobile phone, a headset, a loudspeaker, a hearing aid, a handsfree audio system, a television device, a video recorder, a monitor, a gaming device, a laptop, an audio player, a DVD player, a CD player, a hard disk-based media player, an Internet radio device, a public entertainment device, an MP3 player, a hi-fi system, a vehicle entertainment device, a car entertainment device, a medical communication system, a body-worn device, a speech communication device, a home cinema system, and a music hall system.

17. A method of estimating directivity of a microphone arrangement (110), the method comprising the steps of generating a useful signal (S) by the microphone arrangement (110) in the presence of a useful sound signal source (102) and in the absence of a noise source (201); generating a noise signal (N) by the microphone arrangement (110) in the absence of the useful sound signal source (102) and in the presence of the noise source (201); determining a signal-to-noise ratio (SNR) of the microphone arrangement (110) on the basis of the useful signal (S) and on the basis of the noise signal (N); outputting the determined signal-to-noise ratio (SNR) of the microphone arrangement (110) as a measure of directivity of the microphone arrangement (110).

18. A program element, which, when being executed by a processor (120), is adapted to control or carry out a method of estimating directivity of a microphone arrangement (110) as claimed in claim 17.

19. A computer-readable medium, in which a computer program is stored which, when being executed by a processor (120), is adapted to control or carry out a method of estimating directivity of a microphone arrangement (110) as claimed in claim 17.

20. A method of using a signal-to-noise ratio (SNR) estimated for a microphone arrangement (110) as a measure of directivity of the microphone arrangement (110).