CN101401454A - Stereophonic sound imaging - Google Patents

Abstract

A method for reducing phase differences varying with frequency occurring at certain listening positions with respect to loudspeakers reproducing respective ones of multiple sound channels in a listening space, the phase differences occurring in a sequence of frequency bands in which the phase differences alternate between being predominantly in-phase and predominantly out-of-phase, comprises adjusting the phase in multiple frequency bands in which the multiple sound channels are out-of-phase at such listening positions. Such adjustment of phase includes the frequency bands in which the width of comb filtering pass bands and notches resulting from phase differences at such listening positions would be greater than or commensurate with the critical band width if the phase adjustment were not applied. The listening space may be the interior of a vehicle.

Description

Stereophonic sound imaging

Technical field

The present invention relates to Audio Signal Processing.More particularly, the present invention relates to improve perception acoustic image and the acoustic image direction of using stereo playing system to present, particularly listen under the situation of position with respect to Central Line's symmetry of such stereo playing system at two.Many-side of the present invention comprises equipment, method and is stored in being used on the computer-readable medium makes computer carry out the computer program of described method.

Background technology

The stereophony Play System is almost ubiquitous in many environment, comprises live sound equipment, house music broadcast and automobile audio.To be the sound that sends by a pair of boombox sound different listening to the position with respect to the difference of loud speaker to common effect.These change mainly is to be arrived from each loud speaker by sound to listen to the position and cause listening to the time difference that the position spent altogether on acoustics.The effect of next comprises the interaction in sound and room, but these effects are not discussed here.

The time difference of listening to the position is equivalent to the phase difference with frequency change.For following discussion, term " differential phase between loud speaker " (IDP) is defined as the phase difference that arrives the sound of listening to the position from a pair of boombox.

The time that spends same amount owing to the sound that presents by two loud speakers arrives the ear that is positioned at from two equidistant audiences of loud speaker, (sees Fig. 1 a) so this audience experiences basically less than IDP.From the skew of a pair of boombox, that is, the audience in the place of one of more close loud speaker experiences amplitude along with the IDP that frequency linearity increases (sees Fig. 2 a).

The variation of IDP causes audible not desired effects, comprises the comb filtering of imaging of the audio signal that presents by a pair of boombox and fuzzy.Simple solution is the signal that delay presents by more close loud speaker.Employed retardation makes the signal that presents by two loud speakers arrive audience's ear simultaneously.The result is, be zero for this audience's IDP, and this audience do not experience the imaging pseudomorphism of not expecting.

Yet the use of simple delay is not suitable for the environment such as vehicle, and in vehicle, two audiences can be with respect to a pair of boombox offset from center symmetrically---promptly, and the more close left speaker of audience, the more close right loud speaker (see figure 3) of another audience.Under this environment,, be pickled with grains or in wine so use an audience's of delay correction IDP can cause another audience's experience to become because IDP rate of change on frequency increases.The effect that produces may be factitious, is enough to cause that other audience is significantly uncomfortable.

For directivity and the important audio signal of imaging, that is, have the signal of significant steady-state component, the replacement scheme of time adjustment is directly to adjust IDP,, adjusts the phase place of various frequencies that is.For each frequency, phase place circulates.That is the mapping on phase place to the 360 ° cyclic space of arbitrary value.For this decomposition, phase value is limited to-180 °～180 °, and scope is total up to 360 °.In order to provide circulative example, consider the phase value of 827 ° or 2 * 360+107 °, 827 ° or 2 * 360+107 ° are equal to 107 °.Similarly ,-392 ° or-1 * 360-32 ° is equal to-32 °.Owing to reason discussed below, compare more (promptly with-180 ° or 180 ° near 0 °,-90 °～90 °) frequency be considered to " homophase " or strengthen, the frequency (that is, 90 °～180 ° or 90 °～-180 °) of comparing more approaching-180 ° or 180 ° with 0 ° is considered to (seeing Fig. 4 a and Fig. 4 b) " out-phase " or that offset.

In typical vehicle environmental, each audience's IDP is as follows.0Hz～approximately the frequency between the 250Hz mainly is a homophase---promptly, IDP is between-90 °～90 °.Approximately the frequency between 250Hz～750Hz mainly is an out-phase---promptly, IDP is between 90 °～180 ° or-90 °～-180 °.Approximately the frequency between 750Hz～1250Hz mainly is a homophase.Thisly mainly be homophase and mainly be that the alternating sequence of out-phase increases along with frequency and continues, up to the human ear's of about 20kHz the limit.In this example, every 1kHz repetition of cycle.Frequency band initial frequency and end frequency are the functions of vehicle inside size and audience's position accurately.

Generally acceptedly be, the human auditory system up to about 1500Hz all to the phase difference sensitivity.Thereby below about 1500Hz, the variation of IDP causes the remarkable distortion of the apparent direction in space or the acoustic image of audio signal.This is the distortion except that the amplitude distortion that is caused by comb filtering, and is obtained with above all listening below 1500Hz by the amplitude distortion that comb filtering causes.

Be understood that generally that also the human auditory system is decomposed into a plurality of littler group of frequencies or band group with wide spectrum, is called as critical band.It is poor that critical band represents that two frequencies still can easily be separated the minimum frequency heard, and this difference changes with frequency.At low frequency, critical band is very narrow, along with frequency increases and broadens.In the following discussion, " band " is meant that from a plurality of loud speakers arrival audiences' sound be the frequency band of homophase and out-phase.In the following discussion, critical band is called as " critical band ".

In above-mentioned vehicle environmental, because the peak of about 500Hz and the width of paddy are equal to or greater than CBW, so, can clearly hear the comb filtering effect for the frequency below about 4kHz.More than about 6kHz, critical bandwidth becomes greater than the combined width of a peak and a paddy, and the comb filtering effect becomes and can't hear basically.

Thereby according to an aspect of the present invention, preferably, for the frequency that becomes up to critical bandwidth greater than the aggregate bandwidth of peak of comb filter and a paddy, promptly approximately the following frequency of 6kHz is adjusted IDP.Thereby this can by in two sound channels of audio signal to a plurality of frequency band excute phase adjustment proofread and correct that differential phase realizes between each loud speaker of listening to the position.In case be employed, for two audiences, in the ideal case the IDP that listens to the observed generation in position+/-90 ° in (seeing Figure 11 a and Figure 11 b).Reducing IDP has by this way improved significantly and has perceived as picture, and with amplitude distortion from the complete audible comb filtering with dark and wide null value reduce to+/-the optimum relatively pulsation of 3dB, for most of audiences and sound-content ,+/-the optimum relatively pulsation of 3dB is unheard basically.

Many methods of the prior art only are conceived to the following IDP of about 1kHz.They attempt to proofread and correct the IDP for two audiences in the lowest band, and in lowest band, the sound that arrives the audience mainly is out-phase.They are 180 ° of IDP that add in this band basically like this by using filter and phase shifter.The result is that below 1kHz, the IDP for two audiences after the correction is-90 °～90 °.That is to say that the following frequency of 1kHz mainly is a homophase for each audience, the audience experiences very big improved imaging.The major defect of such method is that they have ignored the more IDP of high frequency treatment, and at high frequency more, phasing may be favourable.

The 4th, 817, No. 162 US patent teaches uses filter and phase shifter to come the frequency in 200Hz～600Hz scope 180 ° of relative phases that add the signal between L channel and the R channel in two sound channels.In this instruction, this frequency range is represented first band, and at first band, it mainly is out-phase (seeing Fig. 5 a and Fig. 5 b) that arrival audience's sound is listened to the position at two.The problem of this instruction is that phase shifter does not provide the phase variation rate at enough fast belt edge place to proofread and correct with the essence that IDP is provided.

The 5th, 033, No. 092 US patent teaches uses filter and phase shifter with 60 °～90 ° and with-60 °～-90 ° in advance in the phase place of another sound channel in advance in the phase place of a sound channel in the frequency range of 200Hz～1kHz.In this instruction, the sound that the 200Hz approximate representation arrives the audience mainly is first band initial of out-phase.When each sound channel in this band during respectively by in advance 90 ° and-90 °, the total relative phase difference in this band is 180 °.Expected result and the 4th, 817, the method for No. 162 United States Patent (USP)s is similar.The important favourable part of this instruction is, because the phase place of each sound channel is adjusted 90 ° at most, so the amplitude distortion in each sound channel is limited to maximum 3dB.Otherwise if produce relative 180 ° phase shift by only a sound channel being carried out filtering, then this sound channel will have audible null value in its amplitude response.That is to say that amplitude response will reduce to zero the transformation between 0 °～180 °, vice versa.

The 6th, 038, No. 323 US patent teaches uses filter and phase shifter with 180 ° of phase places of adding all frequencies more than the 300Hz to.In this instruction, the sound that 300Hz represents to arrive the audience is listened to first band initial that the position mainly is an out-phase for each.Keep out-phase in order to simplify Design of Filter, make the frequency that is higher than first band, the reason of this instruction is human to the IDP of this frequency more than first out-phase band insensitive (seeing Fig. 6 a and Fig. 6 b).Such fact has been ignored in this instruction,, first is with above frequency hereto that is, and the amplitude distortion that is caused by comb filtering can be heard.

Summary of the invention

The objective of the invention is to improve by stereo playing system is the picture that perceives as that is positioned at audio signal that the audience of the central authorities that depart from Play System symmetrically presents.This is by adjusting a plurality of frequency band excute phases in two sound channels of audio signal, and differential phase realizes between each loud speaker of listening to the position thereby proofread and correct.

Description of drawings

Fig. 1 a schematically shows the spatial relationship of listening to position and two loud speakers, and in this spatial relationship, it is equidistant from loud speaker to listen to the position.

Fig. 1 b displayed map 1a equidistant listened to phase difference (IDP) between the desirable ear of all frequencies of position.This example show such IDP that listens to the position as why not with frequency change.

Fig. 2 a schematically shows and listens to the spatial relationship of position with respect to the skew of two loud speakers.

Phase difference (IDP) between the desirable ear of all frequencies of listening to the position of Fig. 2 b displayed map 2a.How this example shows listens to the IDP of position with frequency change.

Fig. 3 schematically shows two spatial relationships of listening to the position, and each is listened to the position and is offset symmetrically about two loud speakers.

Fig. 4 a and Fig. 4 b show that how two for Fig. 3 listen to for each of position IDP with frequency change.

Fig. 5 a and Fig. 5 b are presented at two desirable IDP responses of listening to the position in the system of the instruction of implementing the 4th, 817, No. 162 United States Patent (USP)s.

Fig. 6 a and Fig. 6 b are presented at two desirable IDP responses of listening to the position in the system of the instruction of implementing the 6th, 038, No. 323 United States Patent (USP)s.

Fig. 7 a shows that many-sided feasible realization based on FIR of the present invention is applied to one of two sound channels, is the functional schematic block diagram of L channel in this case.

Fig. 7 b shows that many-sided feasible realization based on FIR of the present invention is applied to one of two sound channels, is the functional schematic block diagram of R channel in this case.

Fig. 8 a is that the filter of Fig. 7 a or the signal of filter function 702 are exported 703 desirable amplitude response.

Fig. 8 b is that the subtracter of Fig. 7 a or the signal of subtracter function 708 are exported 709 desirable amplitude response.

Fig. 9 a is the desirable phase response of the output signal 715 of Fig. 7 a.

Fig. 9 b is the desirable phase response of the output signal 735 of Fig. 7 b.

Fig. 9 c is that (Fig. 7 a) and the desirable phase response of the relative phase difference between 735 (Fig. 7 b) for two output signals 715 of expression.

Figure 10 a shows the tolerance of desirable IDP compensating filter, and it indicates the phase place requirement of its expectation.

Figure 10 b is the phase response with the expectation of the input of accomplishing characteristic filtering device algorithm for design.

Figure 10 c is the weighting function that is used for characteristic filtering device algorithm for design.

Figure 11 a is that the desirable IDP phase response of position is listened on the left side of Fig. 3 when adopting the FIR filter of Fig. 7 a.

Figure 11 b is that the desirable IDP phase response of position is listened on the right side of Fig. 3 when adopting the FIR filter of Fig. 7 b.

The amplitude response of the realization of the FIR filter before Figure 12 display optimization and desirable phase response.

The amplitude response of the realization of the FIR filter of Figure 13 display optimization and desirable phase response.

Figure 14 shows the amplitude and the phase response of the realization of the iir filter that uses the design of group delay method.

Figure 15, Figure 16 and Figure 17 show the phase response to the realization of the characteristic filtering device algorithm for design with different h values.

Figure 18 is the schematic diagram that shows the example of all-pass filter lattice structure realization.

Figure 19 schematically show when a left side, the right loud speaker of neutralization when existing the vehicle front stall listen to position and loudspeaker layout.

Figure 20 shows that schematically aspect of the present invention is applied to the functional block diagram of the structure of Figure 19.

Figure 21 a schematically shows two quadraphony loudspeaker structures of listening to the position that have that can adopt aspect of the present invention.

Figure 21 b schematically shows four quadraphony loudspeaker structures of listening to the position that have that can adopt aspect of the present invention.

Figure 21 c schematically shows four the six sound channels loudspeaker structures of listening to the position that have that can adopt aspect of the present invention.

Figure 22 a and Figure 22 b are the functional block diagrams that its tolerance is presented at the general bank of filters realization of the ideal filter among Figure 10 a.

Figure 23 shows the pole and zero of the realization of the iir filter that uses the design of group delay method.

The iir filter that Figure 24 and Figure 25 show use characteristic Design of Filter algorithm design is before filter order reduces and the pole and zero of realization afterwards.

Figure 26 shows the phase response of the original expectation that is used for characteristic filtering device algorithm for design.

The iir filter that Figure 27 and Figure 28 show use characteristic Design of Filter algorithm design is before filter order reduces and the phase response of realization afterwards.

Figure 29 shows the phase response of the expectation of five pre-distortions after the correction iteration.

Figure 30 shows the phase response of the realization of iir filter after exponent number reduces and proofreaies and correct iteration five times of use characteristic Design of Filter algorithm design.

Embodiment

Fig. 1 a shows the spatial relationship of listening to position and two loud speakers.Listen to position and left speaker d ₁Between distance with listen to position and right loud speaker d ₂Between distance equate.Also shown and represented other equidistant line of listening to the position.Fig. 1 b is presented at phase difference (IDP) between the ear of equidistant all frequencies of listening to the position.In so equidistant position, the perceived direction and the imaging of the content that presents by loud speaker are tending towards nature, as creator of content is desired.

The spatial relationship of position with respect to two loud speakers skews listened in Fig. 2 a demonstration.In this example, listen between position and the left speaker apart from d ₃Less than listen between position and the right loud speaker apart from d ₄Fig. 2 b be presented at listen to the position IDP how with frequency change.Even the IDP dullness reduces, this figure (with all other IDP figure) is presented at-180 °～180 ° equivalence values in the scope.At 0Hz, the signal homophase is turning back to before frequency A is in homophase, and along with increasing frequency, signal moves on to out-phase.This phase cycling repeats along with increasing frequency.The frequency A that cycle take place to repeat directly is associated with the range difference of listening between position and two loud speakers.For example, if apart from left speaker d ₃Distance be 0.75 meter and apart from right loud speaker d ₄Distance be 1.075 meters, then range difference is 0.325 meter.Frequency point A equals speed of sound divided by range difference, and perhaps approximate 330 metre per second (m/s)s are divided by 0.325, and this obtains 1015Hz.Therefore, in this example, the IDP cycle, every 1015Hz repeated.

Fig. 3 shows two spatial relationships of listening to the position, and each is listened to the position and is offset symmetrically about two loud speakers.Fig. 4 a and Fig. 4 b show for two and how listen to for each of position IDP with frequency change.Can find out that for each cycle of IDP, existence mainly is the frequency of homophase and mainly is the frequency of out-phase.That is to say that IDP is-90 °～90 ° a frequency, IDP is between-90 °～-180 ° or the frequency between 90 °～180 °.IDP mainly is that the frequency of out-phase causes the audible effect of not expecting, comprises imaging fuzzy of the audio signal that presents by two loud speakers.

Fig. 5 a and Fig. 5 b are presented at the 4th, 817, and the ideal of the effect of the instruction of describing in No. 162 United States Patent (USP)s is represented.It mainly is the IDP of all frequencies of out-phase that this instruction is added in first frequency band 180 °.In this instruction, the scope of this band is approximate 200Hz～600Hz.Can find out that in Fig. 5 a and Fig. 5 b listen to the position for two, these sound mainly are homophase now.Yet it mainly is the frequency that is higher than 600Hz of out-phase that this instruction has been ignored.The 5th, 033, the instruction of describing in No. 092 United States Patent (USP) is similar to the 4th, 817, and No. 162 United States Patent (USP)s are except handled frequency range is about 200Hz～1kHz.

Fig. 6 a and Fig. 6 b are presented at the 6th, 038, and the ideal of the effect of the instruction of describing in No. 323 United States Patent (USP)s is represented.It more than first vocal cords mainly is the IDP of all frequencies of out-phase that this instruction adds with 180 ° that first vocal cords neutralize to.In this instruction, this is with from about 200Hz.Can find out that in Fig. 6 a and Fig. 6 b the sound in this first band mainly is homophase now.Yet this instruction has also been ignored, and mainly is that the more high frequency band of out-phase makes the band that is in homophase and is in band out of position of out-phase.

According to an aspect of the present invention, mainly be that the IDP of a plurality of frequency bands of out-phase makes in the specific audible comb filtering effect in position of listening to and minimizes by proofreading and correct.Although minimum out-phase frequency band is paid close attention in invention in the past, can realize significant sense of hearing improvement near the IDP of a plurality of bands below the frequency of critical bandwidth by proofreading and correct approximately up to the width of comb filtering passband and paddy.More than the frequency, can not bring the sense of hearing that is embodied as in the picture to improve at this by proofreading and correct out-phase.In vehicle, this frequency is approximately 6kHz, still, changes a little along with the actual inside size of vehicle with to the relative distance of loud speaker.

According to many aspects of the present invention, audio signal is divided into homophase frequency band and out-phase frequency band, and to each out-phase band, 180 ° of phase shifts are added to two relative phases between the sound channel.The optimal way of doing like this is, phase shift is 90 ° in a sound channel, phase shift in another sound channel-90 °.Interchangeable mode is only to be added to a band in the sound channel with 180 °; Yet, the fluctuation that this may cause in the amplitude response of sound channel significantly, not expect.

Embodiment

In many-sided exemplary embodiment of the present invention, one group of filter provides the phase response of the combination phase shift between the sound channel of the alternately band that the amplitude response of substantially flat and generation have 0 ° and 180 °.Fluctuation in the amplitude response of not expecting can give L channel 90 ° of phase shifts, gives R channel-90 ° phase shift (seeing Fig. 9 a, Fig. 9 b and Fig. 9 c).If realize this operation with 180 ° of phase transitions in a sound channel, then in phase transition, amplitude will descend towards-∞ dB.Yet by using only 90 ° of transformations, the maximum in the frequency is fallen (dip) suddenly and is approximately-3dB.More than about 6kHz, phase response is no longer so important, can be set to zero to two sound channel phase responses.

For some Design of Filter, particularly digital filter design, may more efficiently be, but to continue band is carried out phase shift up to nyquist frequency not in the phase shift of frequency termination zone of definition.For other design, may more efficiently be the phase place of the band of the minimum number that the result of only mobile generation expectation is required.Realize for some, may very little or not influence by the quantity of the band of phase shift to effectiveness affects, can determine about by the selection of the quantity of the band of phase shift according to the time smearing of total filter order and generation.

Based on the geometric position described in Fig. 1 a, Fig. 2 a and Fig. 3, the filter response of expectation is a frequency f _dFunction, frequency f _dWith and the left speaker of listening to the position of offset from center corresponding with the wavelength that the path difference between the right loud speaker equates.This is presented in the equation 1:

$f_d=\frac{c}{|d_L-d_R|}$

Wherein, d _LBe the distance from audience to the left speaker, d _RBe the distance from the audience to right loud speaker, c is speed of sound (all distances is unit with rice).

The feature of the phase performance of IDP compensating filter can be described by the tolerance among Figure 10 a, in Figure 10 a, and f _dFor with the corresponding frequency of wavelength that equates with path difference; B is the quantity of band; Δ F _Beg, Δ F _MidWith Δ F _EndBe respectively before first band, between all bands and the width of transition after the last band; Δ P _BndBe phase error with inside; Δ P _Beg, Δ P _MidWith Δ P _EndBe respectively before first band, between all bands and the phase error after the last band.

Though on all are with, these tolerances are appointed as equally substantially, replacedly, can be differently specified them to each band.For example, maybe advantageously, for first band, have very fast transformation, in first band, people's ear is the most responsive to phase place, along with frequency rises, has wideer transformation, to reduce filter order and to improve efficient.

In short, can use the bank of filters that left audio signal and right audio signal are divided into subband to realize filter, in described bank of filters, alternately subband carries out the phase place adjustment, so that the relative phase in these subbands between two sound channels is 180 °.Figure 26 a and Figure 26 b show the example that general bank of filters realizes.May do not needed to postpone to handle, so that any delay coupling that is given is handled in their delay and phase shift by the subband of phase shift.Can be by subband being added up (seeing Fig. 6 a and Fig. 6 b) or realizing reconfiguring of subband by inverse filterbank.

Replacedly, directly designing filter is given the phase response of expectation.

Below in the discussion of finite impulse response (FIR) (FIR) filter, then be based on the example of the design of bank of filters; Yet the bank of filters method can be used infinite impulse response (IIR) filter.After the discussion of FIR filter, discussion can cause many Direct Method of Design of very efficient iir filter.

Finite impulse response filter

Can use finite impulse response (FIR) (FIR) filter and linear phase digital filter or filter function to realize to IDP phase compensation such as the layout in the example of Fig. 3.Such filter or filter function be can design and extremely foreseeable controlled phase place and amplitude response realized.Fig. 7 a and Fig. 7 b show as the block diagram of the of the present invention many-sided feasible realization based on FIR that is applied to one of two sound channels respectively.

In Fig. 7 a example of handling L channel, produce the signal (703 and 709) of two complementary comb filterings, if with these two signals altogether, will have the amplitude response of substantially flat.Fig. 8 a shows the comb filter response of band pass filter or filter function (" BP filter ") 702.Available one or more filter or filter function obtain such response.Fig. 8 b shows the effective comb filter response by the scheme generation of (" delay ") 704 of BP filter 702, time delay or delay function and subtractive combination device 708.In order to make the comb filter response complementary substantially, BP filter 702 should have identical lag characteristic (seeing Fig. 8 a and Fig. 8 b) basically with delay 704.One of signal of comb filtering is carried out the phase place adjustment of 90 ° of phase shifts to expect in the frequency band of expectation.Though in the signal of two comb filterings any can be moved 90 °, in this example, the signal 709 is by phase shift.Selection in the relevant treatment that the selection influence of in the movable signal or another shows in the example of Fig. 7 b, thus the total displacement from the sound channel to the sound channel is as expecting.It only is as a filter of one group of frequency band selection in the example of Fig. 8 a or the signal (703 and 709) that a plurality of filter produces two comb filterings economically that the use of linear phase FIR filter allows to use.Preferably, the delay by BP filter 702 is not with frequency shift.This allows to produce in the following manner complementary signal: primary signal is postponed the identical time quantum of group delay with FIR BP filter 702, and deduct filtered signal (in the subtractive combination device 708 that shows) from the primary signal of delay as Fig. 7 a.Before will be by any frequency that 90 ° of phase shift processes provide constant delay altogether, these delays should be applied to the signal do not adjusted, to guarantee smooth response once more by phase place.

Filtered signal 709 by the 90 ° of phase shifters in broadband or phase shift process (" 90 ° of phase shifts ") 710 to produce signal 711.By having the delay of identical lag characteristic with 90 ° of phase shifts 710 or delay function 712 comes inhibit signal 703 to generate signal 713.In addition summer or summing function 714, the signal 711 of 90 ° of phase shifts and the signal that postpones 713 produce output signal 715.Can use any method in many known methods to realize 90 ° of phase shifts, such as Hilbert transform.Output signal 715 has the gain of basically identical, only have very narrow-3dB to fall suddenly at unmodified band with between by the band of phase shift at the frequency place corresponding to transition point, but output signal 715 has the phase response of frequency change, shown in Fig. 9 a.

Fig. 7 b shows that many aspects of the present invention are applied to another in two sound channels, is the block diagram of R channel in this case.The block diagram of this block diagram and L channel is very similar, and except the signal (being signal 727 in this case) that deducts delay from filtered signal (in this case for signal 723), rather than vice versa.Shown in Fig. 9 b, final output signal 735 has the gain of basically identical, but to had by the frequency band of phase shift-90 ° of phase shifts (with in the L channel shown in Fig. 9 a+90 ° compare).

Show the relative phase difference between two output signals 715 and 715 among Fig. 9 c.Phase difference shows for listen to 180 ° of combination phase shifts that the position mainly is each frequency band of out-phase at each.Thereby the out-phase frequency band mainly is a homophase listening to that the position becomes.What show generation among Figure 11 a and Figure 11 b listens to the correction IDP of position (see figure 3) to each.

FIR amplitude and phase response

Since the character of FIR filter, the impossible FIR filter (except pure delay) that produces all-pass.Thereby, in the filter amplitude response, exist some to depart from inevitably.Realize that for above-mentioned FIR Figure 12 and Figure 13 provide amplitude and the phase response example about two different filter orders.

During the transition region between the band, existence-3dB falls suddenly in amplitude response.Along with filter order increases, the width that falls suddenly diminishes, and accelerates from+/-90 to 0 phase transition.Yet bigger filter order means bigger impulse response.

Though the FIR filter designs easily, they have some characteristic that realizes that many-side of the present invention is not expected.At first, their require that long relatively impulse response realizes requiring amplitude and phase response---long impulse response causes high computation complexity.The second, long impulse response causes taking off the tail effect for the audible time of not expecting of audio signal pulse or that impact.

The consideration that FIR realizes

For efficient, filter among Fig. 7 a and Fig. 7 b or Filtering Processing 702 and 722 are constructed to uniformly-spaced comb filter group respectively, and this is a low pass filter after the comb filter group uniformly-spaced.Can efficiently comb filter be embodied as sparse FIR filter.Can adopt low pass filter to stop at the phase place adjustment of the above band of the cut-off frequency of expection.

Installing or handling 710 and 730 is 90 ° of phase shift filters or Filtering Processing.For the filter that under the sampling rate of 44.1kHz and 48kHz, is suitable for most of audio frequencies, need 400～800 filter taps.Owing to use the embodiment costliness of direct convolution, so fast Fourier transform (FFT) can be used to adopt fast convolution.

In addition, for the sampling rate of 44.1kHz and 48kHz, the low pass filter of Filtering Processing should have 200～400 taps.It also can be made a profit from fast convolution, and can make up with 90 ° of all-pass filters or filter process.

Infinite impulse response filter

Embodiment preferred uses infinite impulse response (IIR) all-pass filter to realize the phase response of expecting.Iir filter has such advantage, that is, for the phase place and the amplitude response of expectation, they typically have than the shorter impulse response of similar FIR filter.Shorter impulse response causes computation complexity to reduce and the time takes off the tail effect and reduces.Yet iir filter is difficult to design.

The group delay method

The iir filter designing technique of most of classics focuses on and specific amplitude response coupling.Yet, have several technology that are used to design the all-pass iir filter.A kind of method that is used for All-pass Filter is based on the minimum p rank of finding the group delay that is fit to expectation.This method can realize by for example using the PC Tools such as MATLAB (MATLAB is The Math Works, the trade mark of Inc.).Can use MATLAB function iirgrpdelay.m, it is the part of Design of Filter tool box.In aspect realization is of the present invention, desirable phase response is the alternately band with sharp-pointed transformation.Because group delay is the first derivative of phase place,, be ± ∞ in transformation place so desirable group delay is 0 in band.Because so discontinuous impossible suitable minimum p order algorithm, so must find the approximation that derivative is not had the desired phase response of discontinuity point.Be chosen as the sine curve of optimally aiming at by phase response, can design iir filter near required response with the band of expectation with expectation.Figure 14 shows the amplitude and the phase response of the filter that uses the design of group delay method.

Yet the group delay algorithm numerical value instability that becomes on bigger exponent number does not restrain usually.In addition, because this algorithm is suitable for group delay, so any error in the group delay causes the more mistake in the phase response that causes owing to integration.Thereby, exist a large amount of trial-and-error methods or parameter search to find filter with expected performance.In addition, because this method only can design little exponent number, so this method may not be suitable for the application that needs the phase place in a large amount of bands to adjust.That is to say that to the range difference of two loud speakers, promptly Δ is apart from big situation.

Characteristic filtering device method

Another technology that is used to design the IIR all-pass filter is a characteristic filtering device method.Referring to, for example following technical papers: T.Q.Nguyen et al, " Eigenfilter Approach for theDesign of Allpass Filters Approximating a Given Phase Response ", IEEE Trans on Signal Processing, vol.42 (9), 09/1994 and Tkacenko etal, " On The Eigenfilter Design Method and Applications:A Tutorial ", IEEE Transactions on Circuits And Systems-II:Analog And DigitalSignal Processing, Vol, 50, No.9, September1994 Http:// www.systems.caltech.edu/EE/Groups / dsp/students/andre/papers/journal/eigen tutorial.pdf.

Approximate least square between the phase response of permission of characteristic filtering device method and expectation fits.Though do not guarantee to generate stable filter, if condition suitably is set, then it produces stable filter reliably.In addition, some alternative manners that have more approaching real least square or more fluctuate near phase place equivalent.Because characteristic filtering device method can numerically stablize, even up to being big filter order, so it is unusual otherwise effective technique.

Characteristic filtering device method is based on finding the error metrics that can be expressed as the filter coefficient quadric form, such as ε=a ^TPa, wherein, ε is an error, and a is the vector of denominator filter coefficient, and P is a matrix.In case be represented as formula, just can use the Rayleigh principle to find a.Thereby the characteristic value of P and error ε are proportional, and the characteristic vector that is associated with minimal eigenvalue is the optimum solution of a.

For all-pass filter, total phase of the filter of exponent number N _H(ω) phase by following formula and denominator _A(ω) relevant:

φ _H(ω)＝-Nω-2φ _A(ω) (2)

Wherein, ω represents with the radian to be the frequency of unit.An approximate evaluation to the least square phase error of all-pass filter is:

$\mathrm{\ϵ}\≈\frac{1}{\mathrm{\π}}\∫W\left(\mathrm{\ω}\right){\left(a^{T}s\left(w\right)\right)}^2\mathrm{d\ω}---\left(3\right)$

Wherein,

s(ω)＝[sin(φ _A，des(ω))sin(φ _A，des(ω)+ω)...sin(φ _A，des(ω)+Nω)] ^T (4)

The weighting that W (ω) provides for the user, φ _{A, des}(ω) be the expectation phase place of denominator.From (1), have

${\mathrm{\φ}}_{A,\mathrm{des}}\left(\mathrm{\ω}\right)=-\frac{1}{2}({\mathrm{\φ}}_{H,\mathrm{des}}\left(\mathrm{\ω}\right)+\mathrm{N\ω})---\left(5\right)$

Next, we can be expressed as quadratic expression with error metrics ε:

ε=a ^TPa, wherein, $P=\frac{1}{\mathrm{\π}}\∫W\left(\mathrm{\ω}\right)s\left(\mathrm{\ω}\right)s^T\left(\mathrm{\ω}\right)\mathrm{d\ω}---\left(6\right)$

Available discrete and come approximate integration:

$P=\frac{1}{\mathrm{\π}}\underset{i=0}{\overset{M}{\mathrm{\Σ}}}W\left(\frac{i}{M}\mathrm{\π}\right)s\left(\frac{i}{M}\mathrm{\π}\right)s^T\left(\frac{i}{M}\mathrm{\π}\right)---\left(7\right)$

Wherein, M is the quantity of the frequency step of cutting apart [0, π].If λ _MinBe the minimal eigenvalue of P, a _MinBe the characteristic of correspondence vector, then Qi Wang filter is:

$H\left(z\right)=\frac{\underset{n=0}{\overset{N}{\mathrm{\Σ}}}{a}_{\min }[N-n]z^{-n}}{\underset{n=0}{\overset{N}{\mathrm{\Σ}}}{a}_{\min }\left[n\right]z^{-n}}---\left(8\right)$

Unfortunately, do not guarantee that the filter that produces is stable.Yet, if adopt following constraint, can find stable filter usually:

φ _H，des(π)＝-Nπ (9)

Characteristic filtering device method Design of Filter

Based on the parametrization that provides among Figure 10 b and Figure 10 c, following formula be can set up and an amplitude that can realize expecting and the filter of phase response produced, so that being provided, proofreaies and correct the IDP that listens to the position.

Provide the phase response of a left side and R channel expectation by following formula:

Provide the least square weight by following formula:

Provide by the quantity B of the band of phase place correction by following formula:

N is the quantity that postpones the corresponding sampling period with relative time:

$n=\frac{|d_L-d_R|}{c}{f}_s---\left(14\right)$

Wherein, f _cFor more than it not with the cut-off frequency of being adjusted by phase place; f _dFor with the corresponding frequency of wavelength that equates with path difference; Δ f _Beg, Δ f _MidWith Δ f _EndBe respectively before first band, between all bands and the width of transition after the last band; ω _Pre, ω _In, ω _OutAnd ω _PostBe respectively be used for before first band, band is inner, between the band and the user-defined weight after the last band; d _LAnd d _RBe distance (is unit with rice) from listening to position to two loud speaker; C is the speed (is unit with m/s) of sound, f _sBe sampling rate (is unit with Hz).

For left filter, in the band of appointment, exist and linear delay between-pi/2 or-90 ° of skews, right filter has+pi/2 or+90 ° of skews.Susceptible of proof also, φ _{H, L, des}And φ _{H, R, des}Satisfy (9), this allows to find reliably stable filter.By selecting different weights, the acutance of may command width of transition and filter order, undulate quantity and transformation.

The characteristic filtering device improves

As described in, can obtain more approaching approximate evaluation by using the iteration weighting function to real least squares error at the paper of T.Q.Nguyen etc.This causes following error metrics:

ε=a _q ^TPa _q, wherein $P=\frac{1}{\mathrm{\π}}\∫W\left(\mathrm{\ω}\right)\frac{s\left(\mathrm{\ω}\right)s^T\left(\mathrm{\ω}\right)}{{a_{q-1}}^{T}c\left(\mathrm{\ω}\right)c^T\left(\mathrm{\ω}\right)a_{q-1}}\mathrm{d\ω}---\left(15\right)$

Wherein, a _qBe filter coefficient the q time iteration; S (ω) is the vector in (3), and

c(ω)＝[cos(φ _A，des(ω))cos(φ _A，des(ω)+ω)...cos(φ _A，des(ω)+Nω)] ^T (16)

Can separate iteration is carried out initialization by what use that the method as among the Tkacenko etc. with the front finds, and can be by the variation ‖ a of the coefficient between the supervision iteration _q-a _Q-1‖ ²With when it is fully little, in the practice about 10 ^-4The time stop to come termination of iterations.Find that this method is the most effective to the design iir filter, it has reduced the pulsation in the filter freguency response significantly.

IIR amplitude and phase response

Characteristic filtering device method with iteration error tolerance is the filter of any exponent number of real estate life reliably.Yet, exist in the tangible performance jump that filter order takes place.

N＝(2h-1)·n，h≥1， (17)

Wherein, n is the quantity that postpones the corresponding sampling period with relative time, and h is an integer.This performance is jumped corresponding with main peak in the desirable impulse response, can come these main peaks of approximate evaluation by using above FIR method to produce very large FIR filter.Integer h finishes to stipulate the maximum quantity of generable flex point in each band.In the practice, it is useful allowing the additional sample of some postcriticals to help minimize the pulsation amplitude, so use following formula in practice:

N＝(2h-1)·n+E，h≥1 (18)

Wherein, E is an additional sample.Find that E=5 provides superperformance.

By design, guarantee that amplitude response is smooth, and realize that it only is because numerical precision causes that any amplitude departs from by the all-pass of appropriate configurationization.Figure 15, Figure 16 show the phase response with different h values with Figure 17.

Iir filter is realized

There are the many filter constructions that are used to realize the all-pass iir filter.Most of basic skills are that filter factor is decomposed into a series of two exponent parts (two second order).If suitably these parts are divided into groups, then this is the good method that realizes general iir filter.Yet,, guarantee that still filter is an all-pass if exist structurally for the application specific architecture of all-pass---coefficient is quantized.This can cause better numerical property, particularly in low precision fixed point is realized.

Owing to the preferred all-pass filter lattice structure of following reason:

1, it is an all-pass structurally, thereby when coefficient was quantized, the result remained all-pass filter.

2, it has good fixed-point performance.Guarantee the lattice coefficient between 0 and 1, intergrade has good overflow attribute.

3, it has structure simply clocklike.Although it has 2 products rather than 1 (all-pass structure of this available direct form is realized), it has the structure of multiply accumulating very clocklike that be transplanted to digital signal processor (DSP) efficiently.

Thereby, in Figure 18, show such realization, that is, and k ₁-k _nBe the lattice coefficient from filter table, x is an input sample, and y is an output sampling.

Can be by using the Levinson recursion based on IIR denominator coefficients a ₁-a _nFind the lattice coefficient k ₁-k _nThis signal flow causes following realization:

a＝x-k[0]＊s[0]；

y＝s[0]+k[0]＊a；

for(i＝1；i<N；++i)

{

a＝a-k[i]＊s[i]；

s[i-1]＝s[i]+k[i]＊a；

}

s[N-1]＝a；

Wherein, a is an accumulator; S is the filter status array; K is the lattice coefficient.

The iir filter exponent number reduces

The minimum p order algorithm of IIR group delay has an advantage that is better than characteristic filtering device method and is that it can design more efficient filter.This is that following (limit in＜6kHz) the zone, in described zone, the phase place of band just is modified because it only uses cut-off frequency.More than the frequency, method for designing is ignored the phase place at higher frequency place at this.Figure 23 shows the pole drawing of the filter that uses the design of group delay method.

Yet,, must adopt constraint φ for the characteristic filtering device method that produces stable filter _{H, des}(π)=-N π (as previously mentioned).When all frequencies of weight 0 being distributed to more than the cut-off frequency, have no idea to guarantee the phase place at π place.Even adopt the zonule also not produce stable filter with near the non-vanishing weight the π.Thereby, the described algorithm pole and zero that distributes equably around the unit circle.This allows filter is the approximately linear phase place, and provides the known phase response to all frequencies.Figure 24 shows the pole drawing of the filter of use characteristic filtered method design.

Find, after characteristic filtering device algorithm produces stable filter, can delete some unnecessary pole and zeros.This can be that cost obtains significant filter order and reduces (up to 75%) with some phase accuracies, as a result of and the filter that produces is the approximately linear phase place at all frequency places no longer.Because the human auditory system is insensitive to phase place at upper frequency, thus tolerable because some phase distortions that the removal of some pole causes, with respect to unaltered filter, these phase distortions will not become and tin obtain.Figure 25 shows the pole drawing of the filter identical with Figure 24, but has removed approximate 73% pole.Figure 27 shows the phase response before reducing, and Figure 28 shows the phase response after reducing.

The effect of the limit of the close unit circle of deletion mainly is the local influence near the frequency it.Yet, will little overall effect be arranged to all frequencies.Therefore, as seen in Figure 28, delete all high frequency poles and can cause from the tangible phase deviation of desired frequency response.

A kind of method of proofreading and correct such phase deviation is that the response of expectation is twisted in advance, uses so pre-distortion in the design of characteristic filtering device.Can find reasonably pre-distortion in the following manner, described mode promptly finds the filter that reduces and the error between the original filter, and deducts this error from the phase response of expectation iteratively.

From equation (10), (11) and (12), given φ _{H, L, des}(ω), φ _{H, R, des}(ω) and W (ω); If eigenfilter is (φ _{H, des}(ω), W (ω) N) comes the function of the filter of design length N for carrying out above-mentioned characteristic filtering device method for designing, establishes eigenfilter_reduced (φ _{H, des}(ω), W (ω), N R) is such function, this function is at first carried out the design of characteristic filtering device, then, by keeping a minimum k limit that exponent number is reduced factor R, wherein, provides k by following formula when according to increase angle arrangement limit:

For the filter of the correction of calculating minimizing, at first find the not minimizing response of left filter and right filter:

a _full，L＝eigenfilter(φ _H，L，des(ω)，W(ω)，N) (20)

a _full，R＝eigenfilter(φ _H，R，des(ω)，W(ω)，N) (21)

Calculate the relative phase between left filter and the right filter:

a _{fullφrel，full}(ω)＝phase(a _full，R)-phase(a _full，L) (22)

Next, carrying out the several times iteration twists in advance with the phase response to the expectation that is sent to characteristic filtering device design example line program.At first, send the initial value of the iteration of phase response with original expectation;

φ _{H，L，des，0}(ω)＝φ _H，L，des(ω) (23)

φ _{H，R，des，0}(ω)＝φ _H，R，des(ω) (24)

For each iterative step i, the filter that reduces based on the RESPONSE CALCULATION of the expectation of upgrading:

a _i，L＝eigenfilter_reduced(φ _{H，L，des，i}(ω)，W(w)，N，R) (25)

a _i，R＝eigenfilter_reduced(φ _{H，R，des，i}(ω)，W(w)，N，R) (26)

And calculate relative phase between left filter and the right filter:

φ _rel，i(ω)＝phase(a _i，R)-phase(a _i，L) (27)

Then, find the filter of current minimizing and the error between the original unbated filter:

Δ _i(ω)＝unwrap(φ _rel，i(ω)-φ _rel，full(ω)) (28)

This error is used to upgrade the response of expectation.Yet, owing to estimate to reduce by above response difference, thus should carry out minimal modification to the response in this scope, though that expectation is avoided is unnecessary discontinuous.To this a kind of method of counting is that frequency from last correction makes the response of expectation become linearity up to Nyquist.

$C\left(\mathrm{\&omega;}\right)=\left\{\begin{array}{cc}{\mathrm{\&Delta;}}_i\left(\mathrm{\&omega;}\right),& 0\&le;\mathrm{\&omega;}\&le;R\&CenterDot;\mathrm{\&pi;}\\ \frac{{-\mathrm{\&Delta;}}_i(R\&CenterDot;\mathrm{\&pi;})}{\mathrm{\&pi;}(1-R)}\mathrm{\&omega;}+\frac{{\mathrm{\&Delta;}}_i(R\&CenterDot;\mathrm{\&pi;})}{1-R},& R\&CenterDot;\mathrm{\&pi;}\&le;\mathrm{\&omega;}\&le;\mathrm{\&pi;}\end{array}\right.---\left(29\right)$

At last, produce the response of the expectation that is used for next iteration.

${\mathrm{\&phi;}}_{H,L,\mathrm{des},i+1}\left(\mathrm{\&omega;}\right)={\mathrm{\&phi;}}_{H,L,\mathrm{des},i}\left(\mathrm{\&omega;}\right)+\frac{C\left(\mathrm{\&omega;}\right)}{2}---\left(30\right)$

${\mathrm{\&phi;}}_{H,R,\mathrm{des},i+1}\left(\mathrm{\&omega;}\right)={\mathrm{\&phi;}}_{H,R,\mathrm{des},i}\left(\mathrm{\&omega;}\right)-\frac{C\left(\mathrm{\&omega;}\right)}{2}---\left(31\right)$

For this method is shown, Figure 26 shows the original phase response to left filter that provides the response that shows among Figure 27 and right filter.As shown in figure 28, after reducing, Response Table reveals significant phase deviation.For correcting action, the phase response of expectation is twisted in advance.Figure 29 is presented at the phase response of five pre-distortions after the iteration.This obtains the phase response of correction in Figure 30.

In the practice, will in eight iteration, greatly improve response.Sometimes, after iterate improvement several times, the result will deviate from the result of expectation, become unstable sometimes.Therefore, it is useful measuring and select the iteration of carrying out the best by the iteration tracking quality.

In vehicle

Fig. 8 (a, b), Fig. 9 (a, b) and Figure 11 (a b) shows filter and the phase response that is approximately 0.33 meter example from each range difference of listening to position to two loud speaker.Thereby first band of being adjusted by phase place begins and finishes at 750Hz at 250Hz respectively, and the every 1kHz of band structure repeats.Though find this example to many vehicle environmental work, can come to be its custom filter by the suitable inside dimension of measuring particular vehicle.

Many vehicles comprise left speaker and right loud speaker (or loudspeaker channel) in the preceding passenger area of vehicle, comprise left speaker sound channel and right loudspeaker channel in the passenger area of back.Because the main sound channel in the past of preceding passenger receives sound, back passenger receives sound from the back sound channel, and because the distance from passenger to the loud speaker can be different for preceding passenger and back passenger, so with using the Δ that is associated with that row's loud speaker and seating position to use realization of the present invention apart from every pair of filter of design is for twice favourable, the preceding loud speaker that the passenger hears before once being used for once is used for the back loud speaker that the back passenger hears.If there is additional row's passenger, each passenger has additional loud speaker, then can repeat realization of the present invention.The result is sitting in the left side of vehicle and every row passenger on right side perceives improved imaging.Be noted that because for from left speaker and the equidistant position of right loud speaker promptly, the passenger IDP that is sitting in every row's seat central authorities no longer is zero, so for the passenger who is sitting in vehicle central, become image degradation.

Multichannel loudspeaker

Many vehicles also use the multichannel loudspeaker system to reproduce FR audio frequency.Typically, woofer is placed on a lower, intermediate frequency/tweeter is placed on an eminence or the front panel.In these multichannel loudspeaker structures, Δ distance common to audience's Δ distance to woofer and to intermediate frequency/tweeter is different.In this case, if crossover frequency is low as to be enough in the frequency range of the band of just being adjusted by phase place, the list that then can not be designed to low frequency and intermediate frequency/tweeter are all worked is to filter.Can improve this problem by many modes.

At first, because the human auditory system is more responsive to phase place than low frequency, thus can be used for Design of Filter to the Δ distance of woofer, and the upper frequency limit of the band of phase place adjustment can be similar to and reduce to the loud speaker crossover frequency.

The second, it is repeatedly right to be produced as every pair of right special filter that separates of low frequency and intermediate frequency/tweeter to use realization of the present invention.By this way, right every pair of low frequency or intermediate frequency/tweeter has the filter of only adjusting the band in the frequency range that drops on loud speaker, and based on, particularly based on loud speaker to audience's Δ apart from every pair of filter of design.

Surround sound

As mentioned above, find that aspect of the present invention is to existing the symmetrical sound quality that presents from the stereophony of listening to the position of axle favourable.Aspect of the present invention also to stereo material have more than two sound channels (such as, multichannel around) have a benefit.Next such application of aspect of the present invention is described.

The quadraphony around

Particularly in automobile market, quadraphony loud speaker is very general.Because common comprises the discrete signal of center loudspeaker around formant, thus make up with left signal and right signal with typically central signal being equal to, and present central signal by left speaker and right loud speaker.Because left speaker and right loud speaker comprise important common content in this case, so aspect of the present invention causes the imaging of central signal content to be improved to the application of left speaker signal and right loudspeaker signal.

Replacedly, before making up, aspect of the present invention only can be applied to central content with left channel signals and right-channel signals.By this way, for the common content that is produced by the center channel signal, imaging is improved, but left signal and right signal do not change.The common content between left audio signal and the right audio signal is seldom or do not have a common content before left audio signal and right audio signal and central content make up for this hypothesis.

The left speaker signal and the right loudspeaker signal that aspect of the present invention are applied to the front are important for transmit described content in correct perceived position.In addition, aspect of the present invention being applied to the back loud speaker also is favourable to listening to experience.For the content of plan from audience back, particularly 6.1 sources (such as Dolby Pro Logic IIx or Dolby Digital EX), the aspect of the present invention that is applied to the back loud speaker assists in ensuring that suitably to make the virtual image of back placed in the middle, and minimizes audible comb filtering effect." Dolby ", " Dolby Digital ", " Dolby Pro Logic ", " Dolby Digital ", " Dolby Pro Logic IIx " and " DolbyDigital EX " are the trade mark of Dolby Laboratories Licensing Corporation.

In vehicle, the directapath between preceding loud speaker and the back passenger is stopped by front stall usually.In order to compensate this, can be in the loud speaker of back with some content mix in the preceding content.By aspect of the present invention being applied to the back loud speaker, can assist passenger's same way as with it serves as that the back passenger is modified into picture.

Five-sound channel around or triple-track LCR present

What Figure 19 showed when having left speaker, center loudspeaker and right loud speaker vehicle front stall listens to position and loudspeaker layout.Point out, center loudspeaker can with left speaker and right loud speaker not on identical axle, but this can postpone adjust by introducing.By this structure, central signal seems the Central Line (between the audience) from vehicle, rather than each audience's front.

A solution to this problem in the past is the rank that is mixed into some signals in the center channel signal in left speaker and the right loud speaker and reduces center loudspeaker in proportion.Because left audience is near left speaker, the close right loud speaker of right audience, so this solution helps to drag central virtual image some to arrive each audience's front.Yet this method is subjected to following true restriction, that is, it also produces the important comb filtering of the central content between left speaker and the right loud speaker.

Find, aspect of the present invention is applied to left speaker signal and right loudspeaker signal improves central virtual image in this loudspeaker arrangement significantly.This is presented among Figure 24.Gain parameter a and b control are mixed into the amount of the synthetic central content in left speaker and the right loud speaker.Can control these parameters like this, so that power conservation.That is to say a ²+ b ²=1.

Six sound channels or seven-channel around

With theater be provided with different, when in vehicle, using six or during seven-channel, they generally include three pairs of loud speakers add possible in before sound channel.In this case, because with above identical, the realization aspect discovery use on every pair of loud speaker is of the present invention is favourable.Common Δ distance can be used for constructing filter or is used to maximize effect, and perhaps each loud speaker row uses the unique filter that calculates apart from unique Δ distance of the nearest audience or the nearest audience of not blocked by seat to having.

Figure 21 a, Figure 21 b, Figure 21 c show three different examples of loud speaker/audience's layout in the vehicle.

Example among Figure 21 a shows to have two quadraphony loudspeaker structures of listening to the position.Since the Δ distance of listening to the position for preceding loud speaker to the back loud speaker to different, so the filter that can use unique design is to handling the signal of every row's loud speaker.

Example among Figure 21 b shows more traditional quadraphony speaker configurations with two row audiences.Because preceding audience mainly listens the loud speaker that sees before, back audience mainly listens and sees below loud speaker, thus since front stall block directivity with loud speaker, do not use many-sided realization of the present invention intrusively and cause arranging at every row and other.In addition, if every row has different Δ distances, then can be every row designing filter uniquely.

Example among Figure 21 c shows three row's loud speakers with two row audiences.As preceding, blocking that front stall provided makes the audience of front mainly listen the loud speaker that sees before.In this example, middle loud speaker and back loud speaker can have the of the present invention many-sided realization that is applied to improve for the back passenger virtual image.Because middle loud speaker and back loud speaker have to back audience's different Δ distance, thus middle loud speaker and after loud speaker each can to have unique filter right.

Implement

The combination of available hardware or software or hardware and software (such as, programmable logic array) realizes the present invention.Except otherwise illustrating, any algorithm that comprises as a part of the present invention is not relevant with any certain computer or miscellaneous equipment inherently.Specifically, can be by according to the instruction here and written program is used various general-purpose machinerys, perhaps can construct more specialized apparatus (such as, integrated circuit) more easily and carry out required method step.Thereby, can realize the present invention on one or more programmable computer system in one or more computer programs of carrying out, each comprises at least one processor, at least one data-storage system (comprising volatibility and nonvolatile memory and/or memory element), at least one input unit or port and at least one output device or port described programmable computer system.Program code is applied to importing data to carry out function as described herein and to produce output information.Can realize that each such program is to communicate by letter with computer system with the computer language (comprising machine, compilation or high-level process, logic OR object oriented programming languages) of expectation.Under any circumstance, described language can be the language of compiling or explanation.

Preferably that each is such computer program be stored in the readable storage medium of universal or special calculating able to programme or device (such as, solid-state memory or medium or magnetizing mediums or optical medium) go up or the computer program that each is such downloads to described storage medium or device, to be used for when the described storage medium of computer system reads or to install when carrying out process as described herein computer is configured and the operational computations machine.Also can consider the present invention is embodied as the computer-readable recording medium of using the computer program structure, wherein, Gou Zao storage medium makes computer system operate to carry out function as described herein in specific predefined mode like this.

Many embodiment of the present invention have been described.Yet, will understand, can under the situation that does not break away from the spirit and scope of the present invention, carry out various modifications.For example, some steps in the step as described herein can independently sort, thereby, can carry out these steps by the order different with described order.

Claims (12)

1, a kind ofly is used to reduce the method for listening to respect to some of loud speaker that the position takes place with the phase difference of frequency change, described loud speaker reproduces each sound channel in a plurality of sound channels in listening space, described phase difference occurs in phase difference and mainly is being homophase and mainly is being that described method comprises in a series of frequency bands that replace between the out-phase:

Adjust the phase place in a plurality of frequency bands, a plurality of sound channels are out-phase in such position of listening in these a plurality of frequency bands.

2, method according to claim 1, wherein, described listening space is a vehicle inside.

3, according to claim 1 or the described method of claim 2, wherein, the phase place of adjusting in a plurality of frequency bands comprises the frequency band that meets the following conditions: if there is not the application phase adjustment, because of the width of the comb filtering passband that produces at such phase difference of listening to the position and paddy will greater than or be equivalent to critical bandwidth.

4,, wherein, there are two sound channels, each sound channel of one or more loudspeaker reproduction according to any one the described method among the claim 1-3.

5, method according to claim 4, wherein, described adjustment is added to two relative phases between the sound channel with 180 ° of phase shifts.

6, method according to claim 5, wherein, with 90 ° of the phase shifts on the sound channel, with the phase shifts in another sound channel-90 °.

7, according to claim 5 or the described method of claim 6, wherein, realize adjusting by one group of filter, this group filter provides the mobile phase response of combinatorial phase response between the sound channel of the alternately band that the amplitude response of substantially flat and generation have 0 ° and 180 °.

8, method according to claim 7, wherein, described filter comprises finite impulse response (FIR) (FIR) filter.

9, according to the described method of claim 7 that is subordinated to claim 5, wherein, described filter comprises infinite impulse response (IIR) filter.

10, method according to claim 9, wherein, the use characteristic filtered method obtains described infinite impulse response filter.

11, be configured to carry out equipment as any one described method among the claim 1-10.

12, a kind of computer program that is stored on the computer-readable medium is used for making computer to carry out as any one described method of claim 1-10.

Images