The invention under all its aspects shall now further be exemplified to the skilled artisan with the help of figures, whereby the above teaching and the further exemplification of the invention opens to the skilled artisan a large variety of realization modes of the present invention.
The figures show:
FIG. 1 simplified, a signal-flow/functional block diagram of a binaural hearing system, which is used for manufacturing the audible signals according to the present invention;
FIG. 2 in a representation in analogy to that of FIG. 1, a first embodiment of the present invention to manufacture the audible signals;
FIG. 3 a schematic top-to-bottom view of individual's head, defining DOA angle with respect thereto and specific areas of space;
FIG. 4(
a) to (d) polar diagram of beam-forming ability of the HRTF and of a cardioid technical beam-former at different frequencies;
FIG. 5 a further embodiment for operating the method according to the present invention and in a representation in analogy to that of FIGS. 1 and 2;
FIG. 6 in a representation in analogy to that of FIG. 5, a part thereof differently operated as a further embodiment of operating the method according to the invention;
FIG. 7 in a representation in analogy to that of FIG. 6, still a further embodiment of operating the method according to the present invention;
FIG. 8 in a simplified functional block/signal-flow diagram, signal processing by alternative transfer functions;
FIG. 9 operating switch-over from signal processing from a first transfer function to a second transfer function according to FIG. 8 and performed according to the present invention under a further aspect;
FIG. 10 in a representation according to that of FIG. 9, a further example of switch-over according to the present invention and with an eye on the present invention under the aspect of audio signal manufacturing, and
FIG. 11 in a simplified functional block/signal-flow diagram of a binaural hearing system as of FIG. 1, the structure thereof when operated according to the present invention under its second aspect, namely of transiting from a first to a second processing mode.
As was already addressed the principal of the present invention is to apply in a binaural hearing system binaural processing only if necessary and to simplify wherever possible such binaural processing so as to reduce over the operating time of the binaural hearing system, power consumption which is especially due to increased processing requirements necessary for binaural signal processing.
In the following detailed description several signal processing modes shall be described, at least a part thereof being selectively activated in dependency of the acoustical surrounding of an individual wearing a binaural hearing system and as e.g. determined by a classifying procedure.
In FIG. 1 there is shown by means of a signal-flow/functional-block diagram a binaural hearing system as addressed by the present invention. An individual I wears a right-ear and a left-ear hearing device 1L and 1R. Each of the hearing devices comprises a respective input converter arrangement 3L and 3R, also referred to as “microphone arrangement” 3L and 3R. The microphone arrangements, in fact acoustical-to-electrical converter arrangements, comprise one or more than one acoustical-to-electrical converters, e.g. microphones. At the outputs A3L and A3R of the microphone arrangements electrical signals S3L and S3R are generated. These signals depend from the input acoustical signals to the microphone arrangements 3L and 3R. If monaural beam-forming is performed at one or both of the hearing devices 1L, 1R, either with a fixed inadaptable beam characteristic or even with a characteristic which may be adapted to the momentarily prevailing acoustic needs, one can assume such beam-forming being performed in the respective microphone arrangements 3L and 3R. Also some signal preprocessing as e.g. amplifying, analog to digital conversion, filtering etc., may be performed in these units. It is nevertheless merely a question of where the delimitation between microphone arrangement and subsequent signal processing is drawn when defining which signal processing is performed in which of the units as drawn in FIG. 1.
Processing signals S3L and S3R input at E5L, E5R to signal processing units 5L and 5R results in signals S5L, S5R at respective outputs A5L and A5R of the signal processing units which are input to respective inputs E7L and E7R of output electrical-to-mechanical converter arrangements 7L and 7R, also referred to as “speaker arrangements” 7L, 7R.
The binaural hearing system of FIG. 1 further comprises a signal transmission link 9 for cross-data-transmission between the hearing devices 1L and 1R. The transmission link 9 may be wireless or wirebound. In the left-ear signal processing unit 5L a signal which is dependent from the input signal S3L is combined with a signal which is dependent from the output signal S3R as transmitted via transmission link 9.
By the complex and normally frequency-dependent weighting factors αL and βL which are controllably variable, the degree of dependency of signal S5L from S3L and S3R is established.
In the signal processing unit 5R, in analogy, a weighted signal combination of S3L and S3R is performed, with respective variable complex and frequency-dependent factors αR and αR.
The weighting factors αL, αR, βL and βR are controlled from the result of classification of the momentarily prevailing acoustical surrounding of the individual I as performed by a classifier unit 11 generating the respective control signals C(α,β).
In FIG. 2 there is schematically shown in a representation in analogy to that of FIG. 1 the binaural hearing system as of FIG. 1 performing signal processing according to the present invention. The input converter arrangement of one of the two hearing devices, according to FIG. 2 of the right-ear device, 3R, is conceived e.g. to have an omni-directional technical beam-forming characteristic. The output signal S3R of the addressed input converter arrangement 3R is applied to the signal processing unit 5R. The input signal S5R to the right-ear speaker arrangement 7R is made dependent from the output signal of the microphone arrangement 3R. The transmission of a signal dependent from the output signal S3L of the left-ear microphone arrangement 3L is substantially disabled as schematically shown in the transmission unit 9 by the open connection.
Thus, the input signal S5R of the right-ear speaker arrangement 7R is practically exclusively dependent from the output signal of the microphone arrangement 3R of the same device 1R.
On the other hand, signal transmission dependent from the output signal S3R of microphone arrangement 3R to the input signal S5L of the speaker arrangement 7L at the left-ear device 1L is enabled as shown in the transmission unit 9 of FIG. 2 by the closed connection.
The weighting factor αL as of FIG. 1 is selected to be approx. zero. Thereby, the input signal S5L becomes practically exclusively dependent from the output signal S3R of the right-ear microphone arrangement 3R.
As explained up to now, signal processing exploits exclusively the microphone arrangement at one of individual's ears to feed the input signals to the speaker arrangements of both ears of the individual.
The signal processing as shown in FIG. 2 is applied whenever the classifier unit 11 detects an acoustical signal source which is located in a specific area F1 as will be explained with the help of FIG. 3.
In FIG. 3 there is schematically shown individual's head I. Direction of arrival DOA is defined counter clockwise with respect to the projection of the sagittal plane SP on a horizontal plane. Signal processing according to FIG. 2 is performed whenever an acoustical signal source is located in a range of DOA
45°≦DOA≦135°
or
225°≦DOA≦315°
and to a limited distance from individual's ear d which may be
0≦d≦0.3 m
Such acoustical situation, where the acoustical source to be perceived is within the spatial area F1 is especially encountered in telephone applications.
The dominating hearing device, as of FIG. 2 the right-ear hearing device 1R, is the ipsi-lateral hearing device. In this acoustical situation, i.e. perception of an acoustical source in the spatial area F1, delaying the signal output at the contra-lateral speaker arrangement with respect to the signal output at the ipsi-lateral speaker arrangement is established by a fixed amount of time.
Thus, the complex weighting factors βL and αR are set to provide for the time delay τo between the respectively output mechanical signals SML, SMR. According to FIG. 2 this is realized by group delay τ provided by a wireless communication link 9 and a delay (τ-τo) provided by setting the weighting factor αR.
We have further mentioned that the ipsi-lateral microphone arrangement, according to FIG. 23R, is or may be operated in the signal processing technique of FIG. 2 with omni-directional beam characteristic. This resides on the following recognition:
In the FIGS. 4a to 4d there is shown the natural beam-forming characteristics of the HRTF at frequencies of 500 Hz, 1 kHz, 2 kHz, 4 kHz. It might be seen that at 2 kHz and above the HRTF provides for a natural beam-forming which is very similar to that of a cardioid-type microphone. The head shadow of the individual provides for an increased amplification on the ipsi-lateral side by about 5 dB compared with the amplification towards the contra-lateral side. Therefore and with an eye on FIG. 2, the HRTF provides for sufficient beam-forming at the ipsi-lateral hearing device, so that no additional technical beam-forming is necessitated at the “primariy”, the ipsi-lateral input converter arrangement.
ILD compensation, which is possibly necessary for optimizing individual's perception, is realized e.g. by respective adjustment of βL, which according to FIG. 2 also provides for the delay τ. Compensation of possible mismatch of the microphone arrangements may not be necessary in the case considered.
Because the contra-lateral speaker arrangement too is fed with a signal which practically exclusively depends from the output signal of the ipsi-lateral microphone arrangement, an improved signal level and signal-to-noise ratio from the ipsi-lateral side—compared to the contra-lateral side—is exploited. The signal processing substantially necessitates only—except possibly control data—a one-directional transmission from the ipsi-lateral to the contra-lateral hearing device with a constant time delay, so that relatively small processing power and supplying power is needed to operate this signal processing mode.
In context with FIG. 2 we have addressed that by establishing the weighting factor αL to be at least approximately zero, the dependency of the input signal to the contra-lateral speaker arrangement 7L from the output signal of the contra-lateral microphone arrangement is substantially disabled. This is not absolutely necessary. It may suffice to significantly reduce the dependency of the input signal to the contra-lateral speaker arrangement from the output signal of the contra-lateral microphone arrangement relative to such dependency from the output signal of the ipsi-lateral microphone arrangement to achieve the advantages as addressed above.
In FIG. 5 there is shown a further mode of signal processing additionally to the signal processing mode as has been discussed in context with the FIGS. 2 to 4.
In FIG. 5 there is shown in a representation in analogy to that of FIG. 1 the binaural hearing system which is controlled to also allow operating in the processing mode as of FIG. 2. The right-ear side of the binaural hearing system of FIG. 5 is again the ipsi-lateral side. The differences to the processing mode as of FIG. 2 are:
The input signal of the ipsi-lateral speaker arrangement 7R is dependent from the output signal S3L of the contra-lateral microphone arrangement too. The weighting factor βR provides for the delay τ as does the weighting coefficient βL. Thereby, signals which originate from the contra-lateral side arrive at the ipsi-lateral side delayed by τo with respect to signals sensed at the ipsi-lateral microphone arrangement, which amount of delay time is again selected to be at least approx. equal to the time amount a signal in the hearable frequency band needs to propagate from one ear along individual's head to the other ear.
The signal which is transmitted over transmission link 9 from the contra-lateral hearing device to the ipsi-lateral hearing device is subtracted from the signal originating from the ipsi-lateral microphone arrangement. This results in binaural beam-forming, whereat a pronounced amplification minimum is established in contra-lateral direction.
Note that in the embodiment of FIG. 5 again the two microphone arrangements 3L, 3R may be selected to be omni-directional. Clearly and if necessary, technical monaural beam-forming abilities may be provided at the two hearing devices, possibly controllably variable as a function of the result of classification in unit 11, so as to further improve signal-to-noise ratio.
The signal processing as shown in FIG. 5, which introduces additional binaural beam-forming ability compared with the embodiment of FIG. 2, is applied there where the acoustical source to be perceived is not anymore in close proximity of the ipsi-lateral ear, so that e.g. additional signal-to-noise improvement is necessary.
This is especially true in the spatial area F2 as shown in FIG. 2, i.e. at distances d larger than 0.3 m.
Whereas for perceiving acoustical sources in the spatial area F2 the dependency of the input signal of the contra-lateral speaker arrangement from the output signal of the contra-lateral microphone arrangement may be substantially disabled, for perceiving acoustical sources outside the DOA according to F1, F2, αL is not minimized towards zero, but signal processing according to FIG. 5 is rather performed symmetrically as shown in FIG. 6. Thereby, this binaural beam-forming mode is only established for source localization if necessary.
As may be seen in FIG. 4 the HRTF as exploited in the embodiment of FIG. 2 to enhance ipsi-lateral source perception provides in the direction of approx. 315 to 330° for the right ear and, in analogy, of approx. 30 to 45° for the left ear, an amplification maximum. In the combination with the embodiment of FIG. 5 there results beam-forming with maximum amplification still for a DOA of 315 to 330° and of 30 to 45°, but with a significantly better attenuation (negative relative amplification) of signals with a DOA of about 180°. This is exploited in the embodiment as shown in FIG. 6, where signal processing is performed mirror-symmetrically, as perfectly clear to the skilled artisan comparing the embodiments of FIG. 5 and of FIG. 6.
Reconsidering the object of performing binaural beam-forming, one of its objects is to establish to the individual the ability to localize acoustical sources. Whereas improving signal-to-noise ratio may be resolved purely by monaural beam-forming, such monaural beam-forming may not establish such ability.
With an eye on FIG. 3 we have noted that binaural signal processing and beam-forming according to FIG. 2 suffices to establish proper source localization, whenever such source is in the area F1 of FIG. 3. Processing power for such binaural beam-forming and thus power consumption are thereby kept low.
Acoustical sources, which are to be perceived in the area F1 are especially sources as occurring in telephone applications.
The adjacent area F2 as of FIG. 3 is served by signal processing as has been shown in FIG. 5. In spite of the fact that, for proper localization of acoustical sources in this area F2, higher technical requirements are to be fulfilled with respect to binaural signal processing and “catching” of the acoustical source, this is achieved by the embodiment of FIG. 5 with relatively small processing power and power consumption.
Still with the target to minimize overall processing power and power consumption for the binaural hearing system during operation, there may be selected a further area of acoustical surrounding where binaural beam-forming may be minimalized, keeping in mind that one of its primary purposes is to allow proper source localization rather than to improve signal-to-noise ratio.
In a further spatial area denoted by F3 in FIG. 3, which may be approximated by a DOA of at most 45° and of at least 315°, there is no need for technical binaural beam-forming, i.e. in this range the two hearing devices 1L and 1R as of FIG. 1 may be operated independently merely with the respective monaural beam-forming for signal-to-noise improvement.
In a possibly remaining spatial area between F1, F2 and F3 signal processing may be performed as has been shown in FIG. 6.
Thereby, in a large percentage of the acoustical surrounding situations a technical, binaural beam-forming signal processing is applied if at all, which is of low processing power and power supply requirement.
The embodiment as shown in FIG. 7, in a representation in analogy to that of FIG. 6, provides for technical binaural beam-forming for stereo enhancement or stereo widening effect. Thereby, there is achieved a binaural beam-forming characteristic approximately showing, in polar representation, an “8” with direction of minimum amplification at zero and 180°.
In all the embodiments of signal processing which have been described it is highly advantageous to exploit the HRTF natural beam-forming ability. Nevertheless, it must be noted that the beam-forming ability of HRTF only starts at frequencies at and above 2 kHz. Thus, it might be advisable to provide the respective monaural cardioid beam-forming for lower frequencies technically. Thus, it might be advisable to provide technical beam-forming which operates e.g. with a cardioid characteristic, up to about 1 kHz and to exploit, for higher frequencies, the HRTF function and its natural beam-forming ability. To do so, in at least two spectral ranges different signal processing is to be established. This may easily be realized once the signal involved is time-domain to frequency-domain converted. Then different spectral components of the addressed signal are easily differently processed. Another aspect which is considered per se inventive is that changeover from one signal processing mode to the other should not cause artifacts to the individual and should thus be performed in a fading manner. This object too may be resolved in an inventive manner upon the respective signal having been transformed from time-domain into frequency-domain.
In FIG. 8 there is schematically shown by means of a signal flow/functional block diagram a method according to a further aspect of the present invention, namely of fadingly switching from one signal dependency mode to another. This technique may nevertheless be applied wherever a signal is to be processed subsequently via different transfer functions and transition shall be controlled.
According to FIG. 8 a signal S to be processed is subjected to a time-domain to frequency-domain conversion as by an FFT unit 20. By the time-domain to frequency-domain conversion the signal is structured in a number of spectral components. The frequency-domain converted signal S is fed to a generic fading unit, in fact generically a filter unit 22. The filter unit 22 is a selective filter bank, whereat the spectral components of the signal S let us say with the components of interest No. 1 to n are grouped in at least two groups, according to FIG. 8 e.g. in three groups Q1, Q2, Q3. The total number of spectral components in the groups Q1 to Q3 is n. By means of a control input C22 the selectivity of filter 22 is varied, i.e. the number of spectral components momentarily assigned to each of the groups, as an example the three groups Q1 to Q3. This shall be exemplified with the help of FIG. 9. In FIG. 9 there is first shown at “S” the spectral components of interest of signal S. At a first point of time t1 all the spectral components of S are assigned to group Q1, group Q2 and Q3 are empty. Controlled by control input C22, in a second moment of time, t2, one of the spectral components, purely as an example, is assigned to group Q2, group Q1 lacks the addressed component and group Q3 is still empty. Still as an example, in a later time moment tx, group Q1 is empty and all the spectral components of signal S are controllably split upon the groups Q2 and Q3.
As shown in FIG. 8 each of the groups Q1 to Q3 is assigned to one output of filter unit 22, which is operationally connected to a specific transfer function, named G1 to G3 in FIG. 8. The three transfer functions G1 to G3 are, according to FIG. 8, assigned to three processing units 241 to 243. The output signals of the processing unit 241 to 243 are summed in a summing unit 26, resulting in a result signal S′. As may be clearly seen by the skilled artisan, in applying FIG. 9 to processing according to FIG. 8 at t1 the overall interesting spectral components 1 to n are processed by G1, resulting in S′ being G1·S.
In moment t2 one of the spectral components is processed in G2, the remaining spectral component still in G1. At tx the signal S is parallel processed frequency-selectively in G2 G3. Now if we consider under a first aspect of the just addressed frequency-selective processing technique beam-forming by the HRTF, which starts to become effective at 2 kHz and as was addressed beam-forming below of 2 kHz by means of technical beam-forming, it becomes most evident that by the structure as shown in FIG. 8 processing G1 will be technical beam-forming for spectral components up to 2 kHz and e.g. G2 will be a transfer function of “unity” for spectral components at and above 2 kHz, thereby to consider the beam-forming ability of the HRTF.
By the control input C22 the sequence in time of spectral components assigned to each of the groups provided is controlled. Thus and with an eye on fading a signal processing from a transfer function G1 steadily to a transfer function G3 without making use of the transfer function G2, it is most evident to the skilled artisan that, with an eye on FIG. 9, first all the spectral components are assigned to group Q1 as shown at moment t1 and finally all the spectral components will be assigned to group Q3, leaving no spectral components left in group Q1. This is shown in FIG. 10, which is absolutely clear to the skilled artisan having understood the sequences in FIG. 9. Further and with an eye on FIG. 10, group G2 may act as an intermediate or temporary group.
Therefrom, it becomes clear that by controlling the group membership of each of the spectral components of a signal to be processed variably in time, as by the control input C22, and assigning to each of the groups different processing transfer functions, overall processing may fadingly be switched from one processing to the other processing mode.
This technique is applied in a good embodiment of the present invention under its first aspect, for fadingly switching between the different signal processing modes as have been described e.g. in context with FIG. 2, FIG. 5, FIG. 6, FIG. 7.
In FIG. 11 there is shown the technique as has been exemplified with the help of the FIGS. 8 to 10 for fadingly switching from signal processing according to FIG. 2 to signal processing according to FIG. 5. The output signals S3L and S3R are time-domain to frequency-domain converted in converter units 20L and 20R, resulting in the frequency-domain signals S3L and S3R. These signals are fed to fading filter units 22L and 22R having respectively, outputs Q1L and Q1R, Q2L and Q2R. A first transfer function G1 accords with the embodiment of FIG. 2 and a second transfer function G2 with the embodiment of signal processing according to FIG. 5. The group Q1 of spectral components is processed by G1, thus according to FIG. 2, the second group of spectral components Q2 by transfer function or processing G2, thus according to the embodiment of FIG. 5. The results of these two processings with different transfer functions G1 and G2 are summed in respective summing units 26L and 26R, the output thereof being frequency-domain to time-domain converted at units 28L and 28R, leading possibly after digital-to-analog conversion to the signals S5L, S5R of FIG. 1. By means of the synchronized control signals C22L and C22R the membership of each of the interesting frequency components to the groups Q1 and Q2 is controlled, so that for switching from processing according to FIG. 2 to processing according to FIG. 5, all interesting frequency components are first members of group Q1 and, staggered over time, are more and more shifted from a membership in group Q1 to membership in group Q2. Fadingly switching over is then terminated when all the frequency components of interest are transferred to be membership of group Q2 and group of Q1 is in fact “empty”.
By the present invention and under a first aspect binaural beam-forming modes have been proposed which are assigned to specific situations of acoustical surrounding and which necessitate little processing power and supply power requirements. On the other hand and under another aspect of the present invention there is proposed to provide at least two processing modes assigned to specific acoustical situations which modes are of relatively small processing power and supply power consumption and between which one may switch system operation.
Still under a further aspect it has been proposed inventively a method for controlled switching from one processing mode to another, which is ideally suited for fadingly switching from one signal processing mode to another according to and in the first and second aspects of the present invention.
Patent Application
20080013762
