CN1141007C - 3D sound regeneration equipment and method for many listeners - Google Patents

Abstract

The present invention relates to a 3D sound regenerating device for a plurality of listeners, which comprises an inverse filter module, a time division multiplexing module and a plurality of loudspeakers, wherein the inverse filter module is used for filtering input sound signals so that listeners can have identical virtual sound sources; the time division multiplexing module is used for sequentially selecting the sound signals filtered by the inverse filter module within a certain interval; the loudspeakers are used for outputting the sound signals selected by the time division multiplexing module into sound. Thus, the 3D sound regenerating device can simultaneously provide identical 3D sound for a plurality of listeners.

Description

Three dimensional sound reclaim equiment and method thereof at a plurality of listeners

The present invention relates to three-dimensional (3D) sound reproduction equipment, especially, relate to the 3D sound reproduction Apparatus for () and method therefor that a plurality of listeners is provided concurrently identical 3D sound.

In the audio frequency industry, people have carried out a lot of effort in the hope of arranging audio amplifier so that regeneration has the sound of complete telepresenc on the point of one dimension or two-dimensional plane.That is all be to be used to regenerate have the signal of sensation when participating in the cintest, in the initial exploitation monophony system in period, stereophonic sound system and up-to-date Dolby Surround sound system afterwards.Yet, be accompanied by the development of multimedia industry, change to signal the regeneration 3D acoustic space about the purpose of the technology of record and regeneration auditory information (being voice signal and visual information) having had the signal of sensation when participating in the cintest from regeneration verily, audio amplifier can be placed on the optional position in such 3D acoustic space.

Today, most of audio frequency apparatus can be realized the reproduction of stereophonic signal rather than monophonic signal.When stereophonic signal was regenerated, that experiences regenerated signal felt range limited installation site in loud speaker when participating in the cintest.Correspondingly, in order to improve sensation scope when participating in the cintest, people are for improving speaker performance and using signal processing to produce virtual signal and carried out a large amount of research.

A canonical system based on above-mentioned result of study is the dolby surround stereo system around renovation process that uses one group of 5 loud speaker.In this system, handle the virtual signal output to rearmounted loud speaker independently, promptly signal is postponed and the loud speaker that the signal that has reduced amplitude is transferred to the back is generated virtual signal according to the space environment of signal.At present, most of family's video cassette recorders and Laser disc machine all use a kind of being called as the technology of " Dolby Pro Logic (Doby specialty logic) " system.Owing to used the equipment of above-mentioned technology, the same sound quality in the perceptual image movie theatre of therefore can regenerating at home.

As mentioned above,, need the loud speaker identical, correspondingly, will produce the problem of cost and installing space with number of channels although can obtain the more faithful to sound of sensation when participating in the cintest by the quantity that increases sound channel.

These problems can be how to listen to and the result of study of feeling to be present in the sound in the 3d space improves by the end user.Particularly, aspect the research of people's otoacoustic emission tonal discrimination, two ears that inventors have studied relevant people greatly influence the problem that source of sound is distinguished in the 3d space.

Above-mentioned research relates to the interaction that is input to two input signals in the ear, that is the sense of hearing intensity difference of the signal amplitude of feeling by auris dextra and left ear that produces by the difference of sense of hearing time in the transmission sound or be input to auris dextra and the difference of the sound phase place of left ear.According to result of study, simulated the characteristic of distinguishing the source of sound that is present in the space point by the people for people's ears.Such distinctive feature is called as head related transfer function (hereinafter referred to as " HRTF ").

HRTF is a filter factor, and this filter factor is used to simulate the path from the source of sound to the eardrum, it is characterized in that having a value that changes along with the relative position between source of sound and the head.In this case, HRTF is described as impulse response or the transfer function in the middle ear of relevant anthropolith, and wherein, signal is sent in two ears when source of sound is present in a point in space.By using HRTF, can carry out sound is sent to from the position of its existence the processing of other optional position the 3d space.

Simultaneously, people also feel that with regard to people's the sense of hearing problem that how can distinguish the 3D acoustic space carried out a large amount of research.To this, proposed a kind of virtual source of sound recently and investigated its practical application.

Usually, have the position on equilateral triangle summit that connects the straight line of two loud speakers as the base, audience's uppick best is stereo.Yet, owing to can not therefore will produce the problem in space listener's position limit at this point.In addition, be unusual difficulty according to the balance of listener's position adjustments sound.

Aiwa (Aiwa) company of Japan has solved this problem by add " single directed " loud speaker that can produce towards listener's strong sound in general loudspeaker apparatus.The maximum characteristics of above-mentioned loud speaker are that the listener can both enjoy the stereo of balance towards any position of loudspeaker apparatus.In general speaker system, if the listener moves to left with respect to loudspeaker apparatus, then the sound that is produced by the right-hand member loud speaker will weaken.Yet because single directional loudspeaker of comprising is 45 ° of inner inclination in the loudspeaker apparatus, so the right-hand member loud speaker will produce a strong sound left and produce a faint sound to the right.Otherwise the single directional loudspeaker in the left end loudspeaker apparatus will produce a faint sound and arrive left end and produce a strong sound arrival right-hand member.Thereby, the sound balance that when the listener is positioned at the left side or right-hand member, produces by two ends, left and right sides loud speaker.

1993, developed a kind of speaker system that virtual reality sound is provided by the JVC of another company of Japan, use this system, can only be arranged in there is not loud speaker in positive loud speaker from reality back uppick sound with two.Above-mentioned loud speaker has used people's auditory properties.The people unconsciously uses the direction of two ear search sound.Transfer voice speed is 340m/s, and the distance between two ears approximately is 20cm.Thereby the time difference maximum of transfer voice to two ear is 1/500s.The difference that arrives the wave volume of two ears is still distinguished an important parameter of audio direction.The information that human obtains from the difference and the eyes of two ears is distinguished the source of sound.Like this, if can control for the time of transfer voice to two ear, then the sound that only produces from two loud speakers just can be full of whole room, makes the listener feel that he is as being sitting in the movie theatre.

Yet the technology of the relevant 3D sound of all that developed is an object with single listener all up to now.That is, current audio reproduction system will provide stereophonic effect when single listener is positioned at its base for the place, equilateral triangle summit of the straight line that connects two loud speakers.Yet, identical and stereophonic effect simultaneously can not be provided under a plurality of listener's situations.

It is very serious that such problem will become under the situation of household audio and video system.As shown in Figure 1, when all kinsfolks sat around around the source of sound, it is good stereo that general household audio and video system can not provide to all kinsfolks.

Recently, the someone has proposed to use more loud speaker to replace two sound channels regeneration that the suggestion in come personally sensation and space is provided by equipping " Doby specialty logic " system again.Yet in above-mentioned system, a plurality of listeners will be limited in the center of the circle that connects each loud speaker and remove to appreciate complete 3D effect.And then, for multichannel audio is provided, the amplifier that corresponding a plurality of loud speaker is provided and drive each loud speaker.Like this, the problem of cost and installing space will be produced.

In order to address the above problem, the purpose of this invention is to provide for a plurality of listeners no matter their position how, can both provide the 3D sound reproduction Apparatus for () and method therefor of identical 3D sound in the identical time to them.

Correspondingly, to achieve these goals, 3D sound reproduction equipment for a plurality of listeners provided by the invention comprises: input signal is carried out filtering so that make each listener can have the inverse filter module of identical virtual source of sound, the time-division multiplex module of a voice signal of having selected by above-mentioned inverse filter module filtered on preset time intervening sequences ground, and by above-mentioned time-division multiplex choice of equipment voice signal be output as a plurality of loud speakers of sound.

Another aspect of the present invention provides a kind of input audio signal and identical 3D sound is offered a plurality of listeners' method by the loud speaker of two or more fixed qties of being used to regenerate, this method may further comprise the steps: (a) the loud speaker transfer function in path between the acquisition above-mentioned loud speaker of simulation and each listener's the ear, (b) by being multiplied each other, effective source of sound transfer function in path between the inverse matrix of loud speaker transfer function and the effective source of sound of simulation and ear of listener obtains the step of filter value, (c) in predetermined interval, sequentially select the step of a filter value, and (d) use selected filter value that input audio signal is carried out process of convolution and the result of process of convolution is outputed to the step of loud speaker.

Above-mentioned purpose of the present invention and advantage will be by with reference to the accompanying drawings to detailed descriptions that preferred embodiment carried out and further show.

Fig. 1 is the key diagram that a plurality of listeners are arranged in the situation of a common three-dimensional regenerative system.

Fig. 2 is the structured flowchart that the 3D sound reproduction system for a plurality of listeners of the present invention is shown.

Fig. 3 be illustrated in one in the Virtual Space source of sound and the dual track regenerative system in concern example between two loud speakers being comprised.

Fig. 4 is the loudspeaker position compensation relationship that illustrates for generating with the virtual source of sound in the dual track regenerative system of transfer function conceptual description.

Concern example between the actual source of sound that Fig. 5 has illustrated the virtual source of sound in the dual track regenerative system and is reversed filtering.

Fig. 6 is the structured flowchart that the loudspeaker position bucking-out system of the Fig. 4 that constitutes with electric-wave filter matrix is shown.

Fig. 7 illustrates the layout and the head of loud speaker that is used for simulating exactly in a plurality of listener positions the test of HRTF.

Describe the present invention below in conjunction with accompanying drawing.

As shown in Figure 2, the 3D sound reproduction equipment at a plurality of listeners of the present invention comprises 100, one time-division multiplex devices 200 of an inverse filter module and a plurality of loud speaker 300.

Inverse filter module 100 is carried out filtering to input audio signal in order to have identical effective source of sound for a plurality of listeners 400, and this module comprises a plurality of inverse filter parts 10,20 and 30.A voice signal in the voice signal that time-division multiplex device 200 has been selected by 100 filtering of inverse filter module is so that meet the predetermined cycle.Loud speaker 300 the voice signal of having selected by time-division multiplex device 200 as voice output.

The HRTF measurement model that needs each position of a plurality of listeners of correspondence in the method for the present invention.This is because compare with the normal place of a listener in the middle of two loud speakers, and it is very big and away from normal place to envision a plurality of listeners' change in location.Thereby, all need accurate HRTF model for loud speaker and each listener.

Employed HRTF is described below among the present invention.

HRTF is the filter factor that obtains by the transmission path of simulating the eardrum from the source of sound to people.In addition, a transfer function on it also refers in showing free space from 1 source of sound to the frequency plane of the transfer voice of people's duct, it also refers to the frequency distortion degree relevant with people's head, auricle and health in addition.

Because the structure of ear, the frequency spectrum of signal will be owing to the irregularly shaped generation distortion of ear before signal reaches duct.Because the direction of amount distortion and sound and distance dependent, therefore, the variation of this frequency content is in that distinguished by the people will be as an important factor aspect the audio direction.HRTF has represented the degree of frequency distortion.

Thereby HRTF is determined by the position of source of sound, and the HRTF of left ear and auris dextra is different mutually for the source of sound of same position.In addition because everyone face and the shape of auricle are all different, so HRTF with each human body difference.

By using the HRTF 3D sound of can regenerating.That is, when HRTF was positioned at specific position and input audio signal carried out process of convolution, sound seemed to produce at specific position.

[formula 1]

y[n]＝h[n]*x[n]＝IFFT{H[k]·X[k]}

As shown in Equation 1, usually, two signal h[n] with x[n] in time domain convolution be carried out two signal H[k that FFT (rapid fourier change) handles] and x[k] IFFT of multiplication in frequency domain (against rapid fourier change) is identical.In advance given HRTF being carried out FFT handles.Generally, select said method to be since in the frequency domain processing speed of multiplication far faster than the convolution algorithm in the time domain.

At first, obtain the corresponding HRTF of original position information with loud speaker, obtain another and the corresponding HRTF of virtual sound source then, and carry out matrix operation.This matrix operation provides the position of loud speaker and the correlation between effective sound source position.Like this owing to can obtain each other relation by matrix operation at the loud speaker of any position, therefore, the quality of regeneration sound can with the location independent of loud speaker.

At first, narration has only a 3D audio reproducing method under the signal listener situation.

As shown in Figure 3, the position of supposing a listener is in the center of the circle that two loud speakers as desired data regeneration 3D sound are coupled together, and then needs 6 HRTF: two HRTF of 4 of two ears HRTFs and two ears from a virtual source of sound to the listener from each loud speaker to the listener.Among Fig. 3, two loud speaker positions separately about L and R represent, Vs represents that the listener wishes from the virtual location of this uppick sound.

Although all sound are actually from two loud speakers and produce, yet the listener feels that sound seems to result from the specific position in the 3d space.By changing the sound self that produces by two loud speakers, input signal and listener to be wished to carry out process of convolution from the HRTF of the specific position of this uppick sound, this can realize.

In order to change the HRTF between two loud speakers and two ears, used an inverse filter.Here, be not transferred to the ear on the left side, and be not transferred to the ear on the right from the signal of right-hand member loud speaker output from the signal of left end loud speaker output.This is the cross interference removing method.After having eliminated the signal that produces by two loud speakers, wish that for the listener HRTF of the direction of uppick sound is carried out process of convolution with input signal.Like this, the listener feels that sound seems not to be from loud speaker but produces from specific position.

With reference to Fig. 4, square frame C110 is the electric-wave filter matrix that is used to simulate the transfer voice path of two ears from two loud speakers to the people, and to be simulation wish the electric-wave filter matrix in the path of the sound two ears of uppick from the listener of virtual source of sound transmission to square frame D120.Square frame H130 is used for inverse filter matrix that the relation between virtual source of sound and two loud speakers that are set up is compensated, wherein, before outputing to loud speaker input signal is carried out process of convolution.Fig. 5 shows the notion with co-relation.

Fig. 6 shows the computational methods of inverse filter H.That is, when two input signals are L and R respectively, be transferred to the final output signal Y of two ears from loud speaker _LAnd Y _RCan be described below:

[formula 2] $\left[\begin{array}{c}{Y}_L\\ Y_R\end{array}\right]=\begin{array}{c}\left[\begin{array}{cc}{C}_{\mathrm{LL}}& C_{\mathrm{RL}}\\ C_{\mathrm{LR}}& C_{\mathrm{RR}}\end{array}\right].\left[\begin{array}{cc}{H}_{\mathrm{LL}}& H_{\mathrm{RL}}\\ H_{\mathrm{LR}}& H_{\mathrm{RR}}\end{array}\right]\end{array}.\left[\begin{array}{c}L\\ R\end{array}\right]$

In addition, cheat and decide the listener and wish that the virtual output valve from the position of this uppick sound is V ₁And V _R, then can be described below with co-relation:

[formula 3] $\left[\begin{array}{c}{V}_L\\ V_R\end{array}\right]=\left[\begin{array}{cc}{D}_{\mathrm{LL}}& D_{\mathrm{RL}}\\ D_{\mathrm{LR}}& D_{\mathrm{RR}}\end{array}\right].\left[\begin{array}{c}L\\ R\end{array}\right]$

As a result of, under desirable state, formula 2 and formula 3 are with identical.In fact, if the difference between two formula is very little then consistency is fairly good.Suppose that above two formula are identical, then inverse filter H matrix can obtain from following formula.

[formula 4] $\left[\begin{array}{cc}{H}_{\mathrm{LL}}& H_{\mathrm{RL}}\\ H_{\mathrm{LR}}& H_{\mathrm{RR}}\end{array}\right]={\left[\begin{array}{cc}{C}_{\mathrm{LL}}& C_{\mathrm{RL}}\\ C_{\mathrm{LR}}& C_{\mathrm{RR}}\end{array}\right].}^{-1}\left[\begin{array}{cc}{D}_{\mathrm{LL}}& D_{\mathrm{RL}}\\ D_{\mathrm{LR}}& D_{\mathrm{RR}}\end{array}\right]$ $=\frac{1}{C_{\mathrm{LL}}{C}_{\mathrm{RR}}-C_{\mathrm{LR}}{C}_{\mathrm{RL}}}{\left[\begin{array}{cc}{C}_{\mathrm{RR}}& -C_{\mathrm{RL}}\\ {-C}_{\mathrm{LR}}& C_{\mathrm{LL}}\end{array}\right]}^1.\left[\begin{array}{cc}{D}_{\mathrm{LL}}& D_{\mathrm{RL}}\\ D_{\mathrm{LR}}& D_{\mathrm{RR}}\end{array}\right]$

Below explanation is for a plurality of listeners' renovation process.

In for the renovation process under a plurality of listener's situations, an accurate HRTF model corresponding to each tin receipts person's position should be proposed at first.Because typical HRTF modeling is transfer function when a listener is positioned at the center as the Kemar that provides with MIT, so in fact it can not be applied among the present invention.Thereby, measure the HRTF that meets listener positions, must arrange testing equipment as shown in Figure 7.Here, the distance setting between each listener is that the position of 30cm and two loud speakers is respectively departed from 30 ° to left and right, and this is the stereophonic reproduction position of a standard.By using the HRTF of above-mentioned each listener positions that obtains, recomputate each inverse filter, so that obtain and the corresponding inverse filter module 100 that comprises a plurality of inverse filter parts 10,20 and 30 of each listener.

The time separating method of core of the present invention below will be described.

Sequentially select the inverse filter independently handled for each listener at a certain time interval, and by two loud speakers regeneration by selecteed inverse filter section processes voice signal.Although because the scene that actual switching is formed and discontinuous, yet because persistence of vision phenomenon, our eyes feel to seem an interior at certain time intervals continuous continuous scene, thereby said method is feasible.Promptly, although the result of each filter process is separate with each listener's position, but when this result sequentially outputed to loud speaker in the preset time interval, each listener can both feel that he has heard continuous sound in his/her position.

Here, most important thing is the recovery time interval for each position.If the recovery time for certain position is provided with longly at interval, then other listeners in other position listen to less than sound.In addition, if the recovery time is at interval too short, then the listener can not have adequate time and removes to listen to a complete sound.

Effect of the present invention is as follows.

For the voice signal of being imported by two loud speaker regeneration, making provides identical 3D sound for a plurality of listeners, obtain the loud speaker transfer function to the path two ears of each listener's simulation from two loud speakers to each listener.Here, listener's position can be defined in the special scope, and needn't be limited to the center.

In addition, the inverse matrix of virtual source of sound transfer function by path between the ear of simulating virtual source of sound and listener and loud speaker transfer function multiplies each other and obtains filter value.Input audio signal carries out process of convolution with a filter value.

Select a filter value and output to loud speaker on predetermined intervening sequences ground.Owing to need at least the time interval of 20ms to remove to distinguish sound for the listener, thus among the present invention the regeneration of each listener positions at interval at least greater than 20ms.In addition, if a large amount of listeners is arranged, then remove to handle signal owing to will take a lot of times for all listeners, therefore there is restriction in time-division multiplex method of the present invention at listener's quantitative aspects.

In the embodiment of the invention of being introduced, can adjust changeably according to listener's entire quantity and construct the time-division at interval.

In addition, more than the restricted number of loud speaker is two in the narration, yet the present invention also can use a plurality of loud speakers.Thereby, it should be noted that the present invention is not limited to the foregoing description, in spirit of the present invention and the scope determined by additional claim, can produce variation and the change of carrying out with known technology.

As mentioned above, according to the present invention, then can only use two loud speakers just to appreciate 3D sound and can simultaneously offer a plurality of listeners to identical 3D sound effect.

Particularly, when in the room, appreciating the audio/video household audio and video system, freely be sitting in the identical 3D sound of loud speaker all kinsfolks before uppick simultaneously and appreciate sound true to nature.

Claims (3)

1. 3D sound reproduction equipment at a plurality of listeners is characterized in that comprising:

The inverse filter module is used for that input audio signal is carried out filtering and makes each listener can have identical virtual source of sound;

The time-division multiplex module, be used for predetermined intervening sequences ground selection by above-mentioned inverse filter module filtered a voice signal;

A plurality of loud speakers, be used for by above-mentioned time-division multiplex choice of equipment voice signal as voice output;

Above-mentioned inverse filter module comprises a plurality of inverse filter parts identical with listener's quantity, each above-mentioned filter segment all has a filtering characteristic, and this filtering characteristic is by the above-mentioned loud speaker of simulation with corresponding to the inverse matrix C of the loud speaker transfer function C in path between the listener's of inverse filter part the ear ^-1And simulate the value that the virtual source of sound transfer function D in path multiplies each other and obtains between a virtual source of sound and the listener's ear.

2. the loud speaker by two or more fixed qties is regenerated input audio signal so that provide identical 3D sound method to a plurality of listeners, the method is characterized in that may further comprise the steps:

(a) obtain the loud speaker transfer function, this transfer function is simulated the path between above-mentioned loud speaker and each listener's the ear;

(b) obtain filter value, this filter value is by acquisition that the virtual source of sound transfer function in path between the inverse matrix of loud speaker transfer function and the virtual source of sound of simulation and ear of listener is multiplied each other;

(c) in effectively at interval, sequentially select the step of a filter value;

(d) use selecteed filter value that input audio signal is carried out process of convolution and the result of process of convolution is outputed to loud speaker.

3. as the method for record in the claim 2, it is characterized in that:

Above-mentioned predetermined space in the above-mentioned steps (c) and above-mentioned listener's quantity change pro rata.

Images