The invention concerns a method for eliminating room modes that are created as resonances in a room when a user signal is played through a main loudspeaker.
In every room, e.g. in recording studios, resonances form at certain frequencies, so-called room modes. In comparison to other sounds played at the same level, these are distorted in level and more persistent and are perceived as disturbing.
A room mode can form when a half wavelength of a tone or an integer multiple thereof corresponds exactly to the length, width or another important dimension of the room in which the tone is generated. For comparison, we can consider the excitation of a guitar string, in which root notes and overtones are formed. The frequencies of the disturbing room modes are in the bass range, with higher overtones being less pronounced.
The intention is that the original sounds of a played user signal are heard completely and correctly in all frequencies and then fade away as if the room were infinitely large. No sounds from the original signal should be suppressed, boosted or temporally altered. Sounds reflected from the walls of the room are permitted per se, except in the frequencies of the room modes.
Known solutions to this problem use a filter that equalises the volume of the sound output for problematic sounds according to the problem. However, these solutions change the original sound of the sound signal to be emitted, which is why they should be avoided.
In U.S. Pat. No. 10,490,180 or WO 2017/037341 A1, from first digital information that describes a system impulse response caused by a loudspeaker, second digital information is derived that describes a cancelling impulse response and is added attenuated and with a time delay. For this purpose, a delay value must be determined.
The disadvantage of this or any other method that determines a delay value is the limitation that the correction response generated in this way is only designed for a few specific room modes which can be eliminated. However, a large number of different room modes form in each room, for example in the longitudinal and transverse directions, in height, but also diagonally. Depending on the frequencies, different room modes are dominant in each case. Each room mode forms its impulse response with a different, unique delay time. The decisive factors are the type of room mode and the position of the loudspeaker and microphone. Naturally, only a small subset of all modes can be eliminated with a correction signal specified with a time delay.
In EP 2357847, a delay response that is to be equalised is assigned for a loudspeaker at a predefined position within a listening room. Filter coefficients are then calculated for all-pass filters that are arranged respectively upstream of this loudspeaker, wherein the all-pass filter has a transfer characteristic such that the corresponding delay response corresponds to a predefined target delay.
These solutions usually have the disadvantage that a precise time delay must be determined so that the Active Noise Cancelling/Control signal (ANC signal) is superimposed on the original signal in the correct phase position. In addition, the computing power is very high because the ANC signal must be calculated in a very short time and provided with high resolution in order to be able to output the correction signal in time.
The object of the present invention is therefore to describe a method which can play the complete original sound sequence of a specified item in a room and at the same time prevent the formation of the room modes in the room. This should preferably be possible with relatively low computing power so that timely correction of the original signal can take place using commercially available processors, such as those commonly used in consumer goods, for example.
Further tasks of the invention are to describe a digital signal processor (DSP) with which this method can be executed, as well as a loudspeaker that can fill the room with the user signal or with a correction signal, without room modes disturbing the acoustics in the room.
The tasks are solved by the claims in the corresponding categories. Advantageous methods and embodiments are described in the respective dependent claims.
The embodiments according to the invention are described in connection with the figures.
The solution of the invention essentially consists in the fact that in a two-stage characteristic value measurement, a filter W*(z) is generated and saved with the aid of a digital signal processor, this filter characterising and mapping the sound changes of a user signal N emitted by a main loudspeaker into a room, including the room modes Ñ generated in this way. From the available digital user signal N which is played by the main loudspeaker, the filter W generates a modified signal Ñ− to eliminate the room modes Ñ, this signal being played by a correction loudspeaker. The two signals cancel each other out in the room. Since the passage through the filter W requires a certain time dt, the original signal N remains fully audible in the room and cannot be eliminated. The signal Ñ− can therefore only cancel out that portion of the sound waves in the room still present after this time dt. This ensures that the original signal N is fully audible in the room, while the persistent room modes Ñ are cancelled out after a short time dt. The main loudspeaker and the correction loudspeaker can be one and the same loudspeaker.
The user signal N must be present as a digital signal or be digitised from an analogue signal so that the method according to the invention can be carried out with it.
In the first stage of the characteristic value measurement, the acoustic characteristics of a digital signal A are reproduced and saved in an intermediate filter S′*(z) after that signal has been played by the correction loudspeaker in this room. For this purpose, the signal B is received and analysed at a microphone in the room.
In the second stage, a signal C is played into the room from both the main loudspeaker and the correction loudspeaker and is received by the microphone as signal D. In addition, a filter W(z) is generated which changes the signal C before it is sent to the correction loudspeaker. The filter W is now changed and adjusted until the sound at the microphone is cancelled out as much as possible. This is possible because the characteristics of the correction loudspeaker are already known as S′*(z). This filter is finally saved as W*(z) and used for the utilisation of the method.
The invention will be explained in more detail below in conjunction with the drawings. The figures show the following:
For the method according to the invention, a main loudspeaker 21 and a correction loudspeaker 22 must be arranged in the room 10, wherein these can be designed as separate loudspeakers or, alternatively, as a common loudspeaker 23. In addition, a microphone 30 must be arranged in the room, with the loudspeaker or loudspeakers 21, 22, 23 and the microphone 30 being connected with appropriate sound electronics to a digital signal processor 60.
Only one loudspeaker 21, 22, 23 is shown in each case, but there can also be several loudspeakers 21, 22, 23 with the same functions, which are likewise connected to the necessary sound electronics 40.
With the device shown in
In principle, the method can be carried out using the two options mentioned here. In the first variant, a main loudspeaker 21 and a separate correction loudspeaker 22 are provided; in the second, these are designed together as a common loudspeaker 23. The first variant is described in more detail in
The method according to the invention essentially comprises a first and a second characteristic value measurement, shown in
The invention is described in detail in the following:
The method according to the invention serves to eliminate room modes Ñ which form as resonances in a room 10 when a digital user signal N is played by a main loudspeaker 21. It is characterised by the following steps:
For elucidation, accordingly first of all the room 10 is set up in step a. Care should be taken here to ensure that the loudspeaker or loudspeakers 21, 22, 23 always remain in the same place. The characteristic value measurements only need to be carried out once in each case. Any user signals N can then be played until the room geometry has changed in relation to the loudspeaker positions.
Thus if, for example, a room is divided or the loudspeaker position is changed, new characteristic value measurements are necessary. The position of the loudspeakers 21, 22, 23 should therefore be chosen carefully.
Since room modes Ñ are usually low frequency, it is advisable to use subwoofers as loudspeakers 21, 22, 23, as these are suitable for playing low frequency sounds.
In principle, the microphone 30 can be arranged at any location in the room 30. However, it has proven advantageous for the microphone 30 in step a to be positioned at a location either where a person 11 is likely to be in step b, or near a wall 13 of the room 10 that is far away from the main loudspeaker 21.
The aim of the first characteristic value measurement, shown in
The reproduction is adjusted with the aid of the LMS module and takes approximately between 10 and 30 seconds; then the determined changeable filter S′(z) can be saved as an unchangeable filter S′*(z). It is needed only for the second characteristic value measurement but is not used for the later usage phase. Any signal can be used as signal A that has a sufficient frequency content of all relevant low frequencies. White or, preferably, pink noise has proven to be useful.
Subsequently, the signal processor 60 can be set up for this second characteristic value measurement. The changeable filter W(z) is now at the location of the previous changeable filter S′(z). Its output leads to the second or common output 63, 64 and finally to the correction loudspeaker 22 or to the common loudspeaker 23, as shown in
The aim of this second characteristic value measurement, shown in
It should be noted that through the use of the filter W(z), the signal D which is forwarded by the microphone 30 and which originates from the sound waves 24 from the primary path P, CP(z), is soon superimposed by the sound waves 24 from the secondary path S and is therefore modified. The best possible cancellation of the signal D is achieved when the sound waves caused by the correction loudspeaker 22 compensate as far as possible for those caused by the main loudspeaker 21. This in turn means that the signal C on the primary path P changes as similarly as possible to the way through the filter W(z) and subsequently via the secondary path S. Since this secondary path S is already known from the first characteristic value measurement in the form of S′*(z), the resulting filter W*(z) can be determined in the same way as S′*(z) was previously. White or, preferably, pink noise can again be used as the output signal C; the process takes a similar amount of time to the first characteristic value measurement.
However, the LMS module can only recognise and react to causal relationships in the signals from its two inputs 71, 72. It changes the electronic filter W(z) in such a way that there are no longer any components in the signal D that have a causal relationship with the signal from the filter position 74. Accordingly, for example, ambient noise that only enters the LMS module via the input 72 is irrelevant and has no influence on the characteristic value measurement or on changing the electronic filter W(z).
The passage of the signal C through the electronic filter W(z) takes a certain amount of time dt. The signal in the secondary path S is therefore emitted correspondingly later by the correction loudspeaker 22, 23 than in the primary path P by the main loudspeaker 21, 23. Thus, the signal in the secondary path S cannot eliminate the original signal C in the primary path P because it is too late. The room modes Ñ, on the other hand, continue to oscillate for a long time in the room 10, and spread there. They have a causal connection with the original signal that caused them and can therefore be eliminated by the correction loudspeaker 22, 23.
The signal C needs a time dt to pass through the filter W(z) and also a time dts to pass through the secondary path, with dt being much shorter than dts under normal conditions. In addition, the signal C needs the time dtp to pass through the primary path. The arrangement of the loudspeakers in
It should be noted that none of the times dt, dts or dtp are determined or known. They are therefore not included in the method. The time delay dt of the delayed playback of the correction loudspeaker 22 corresponds to the time that the user signal N needs to pass through the filter W*(z). This value is determined solely by the filter W*(z). There is also no need to enter a delay in the system or in the process in order to intentionally send the signal Ñ− into the secondary path S later.
Preferably, a common loudspeaker 23 is used, to which the two signals N, Ñ− which were provided individually for the main loudspeaker 21 and the correction loudspeaker 22 are fed in a superimposed manner. This is certainly advantageous for cost reasons. In this case, the digital signal processor 60 can be arranged directly in the common loudspeaker 23, preferably integrated.
However, if a loudspeaker is already present that is to continue to be used, an additional correction loudspeaker 22 can be used. This is preferably a subwoofer, which preferably has similar acoustic properties to the existing loudspeaker since it hardly needs to play any high frequencies. The digital signal processor 60 and possibly other components of the sound electronics 40 can then likewise be arranged, preferably integrated, in this additional correction loudspeaker 22.
A microphone 30 is no longer required for using the method. This is preferably now detached and removed. The loudspeaker or loudspeakers 21, 22, 23 must however remain in their places and for utilisation too they are connected to the signal processor 60 with the sound electronics 40 required for this. Equivalent products should be used for the sound electronics 40 as for the characteristic value measurements, with characteristics as similar as possible.
In this phase of the method, only the components of the signal processor 60 are used, as shown in
When used in the signal processor 60, a user signal N is simultaneously routed on the one hand to the first or common output 62, 64 and on the other hand through the filter W*(z), and subsequently as signal Ñ− to the second or common output 63, 64. Accordingly, the signal N from the main or common loudspeaker 21, 23 is played somewhat earlier than the signal Ñ− from the correction or common loudspeaker 22, 23, namely by the time dt which is needed for passing through the filter W*(z).
During this time dt, room modes N form in the room 10, which are eliminated by the correction loudspeaker 22 played with a delay. The original signal remains completely audible here.
The method does not delete sounds that originate from sources other than the digital user signal. The filter W*(z) can react to and neutralise only those sound waves that have a causal connection with the original signal and that are still present after the user signal N has passed through the filter.
Only those acoustic sound waves are deleted that have a causal connection with the user signal N and are still present in the room 10 after the time delay dt. These are the room modes Ñ.
In the following, the digital signal processor 60 according to the invention is described here with the aid of
A digital signal processor 60 according to the invention for use in a method described above comprises:
Other, modified signal circuit diagrams are also suitable if they can be used to carry out the methods according to the invention.
With these two variants of the signal processor 60 shown here, both the characteristic value measurements and the use of the method can be carried out. For this purpose, a first switch 77 is arranged, and in the variant according to
The first switch 77 is arranged at the output of the filter 70, which can be occupied with S′(z), W(z) or W*(z), and can lead optionally to the subtractor 75 or to the second or common output 63, 64. For the first characteristic value measurement, the first switch 77 is connected to the subtractor 75 so that the signal from the output of the filter 70 is subtracted from the microphone signal B which is received at the microphone input 65. After that, for the second characteristic value measurement, and for using the method, the first switch 77 will lead the signal from the filter 70 to the second or common output 63, 64.
In the second characteristic value measurement, the subtractor 75 does not receive a second signal for subtraction and accordingly passes the signal D unchanged to the LMS module. As is known, no microphone is used when using the method. An interrupter 79 can therefore be provided after the microphone input 65 in order to avoid interference.
Additional interrupters 79 can be arranged before or after the filter position 74 and/or at the control output 73 of the LMS module. They can all interrupt the lines during use of the method. However, they must ensure a connection during the characteristic value measurements. The interrupters 79 are optional and can also be omitted.
A further switch 78, see
As already described, interrupters 79 can also be arranged here which can interrupt the connections to and from the LMS module. If these are not arranged, the LMS module must be prevented in another way from exerting an influence on the unchangeable filter W*(z).
In summary, after the first characteristic value measurement, switch 77 and, if necessary, switch 78 must be switched over, and after the second characteristic value measurement, the microphone can be removed and the LMS module with the upstream filter position 74 can be uncoupled.
As soon as the room or the position of the loudspeakers 21, 22, 23 are changed, the first and second characteristic value measurements can be carried out again. To do this, the switches 77, 78 must be set accordingly again and the LMS module with the upstream filter position 74 must be connected. In addition, the microphone 30 must be set up and connected again.
In a preferred embodiment, the signal processor may comprise a sound generator 50 for performing the characteristic value measurements, wherein the sound generator can preferably generate white or pink noise.
According to the invention, a loudspeaker 21, 22, 23 is also described here which comprises a digital signal processor 60 according to the invention, wherein it is preferably a subwoofer.
This loudspeaker is preferably the correction loudspeaker 22 or the common loudspeaker 23.
Number | Date | Country | Kind |
---|---|---|---|
CH070572/2021 | Nov 2021 | CH | national |
Filing Document | Filing Date | Country | Kind |
---|---|---|---|
PCT/EP2022/081803 | 11/14/2022 | WO |