Automatic-threshold group compression method and multichannel compressor

Abstract

The invention relates to a multichannel compression method with automatic threshold definition for a group of audio signals, comprising the following steps: feeding the audio signals into individual audio compression channels, wherein each of the audio compression channels has a controllable output amplifier; ascertaining the individual gains for the output amplifiers of the individual audio compression channels by forming a sum signal and ascertaining a sum level based on the sum signal, and implementing control signals, representing the gains, for the individual audio compression channels based on a comparison between the sum level and an adaptively determined threshold, wherein a channel level is determined for each channel and a current maximum channel level is determined based on the channel levels, and wherein the current maximum channel level increased by a positive offset is determined as the adaptive threshold. The invention furthermore relates to a compressor for carrying out the method.

Description

[0001] Applicant: STAGETEC GmbH 12.09.2025 Our Ref: P19.090WO Method for Group Compression with Automatic Threshold and Multi-Channel Compressor The invention relates to a method for group compression of several audio signals and a multi-channel compressor. It is known to use mixing consoles in audio signal processing that have a multitude of inputs for audio signals. In a mixing process, a multitude of audio tracks are fed to a mixing console. The audio signals from the audio tracks undergo a multitude of different signal processing steps during the mixing process and are finally combined into a single- or multi-track output signal with a specific number of audio tracks, for example in a final format such as mono, stereo, or surround.In this general process, it is common practice to combine several related audio tracks, such as those of a miked drum kit consisting of the audio tracks kick, snare, overheads, toms, and hi-hats, into a group signal, also called a bus signal, i.e., to summe them into a separate unit or "group." The various audio tracks are often named after the instruments whose recorded audio signals they represent. In a mixing console, the components intended for processing an audio track or signal are also called channels or channel strips. Similarly, the various audio signals processed in a mixing console are also referred to as channels, with the term "channel" sometimes being used synonymously with the corresponding audio signal of that channel.The process of assigning an input source to a channel strip, or a channel strip to a summing bus or direct output, is also known as routing. Multiple channel strips or direct sources can be routed to a common bus by summing their weighted signals. Such a summed bus signal, or group signal, can then be subjected as a whole to subsequent steps, such as collective compression. Compression, in the context of audio signal processing, serves the purpose of limiting the dynamic range of an audio signal. This is achieved through controlled attenuation of the audio signal. Such compression of a group signal, such as that of the drum track, leads to a limitation of the dynamic range of the drum signal. (Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO)In this process, the group signal often undergoes a subjectively perceived compression or blending, which is a desired effect. An adjustable gain boost can optionally compensate for the level loss caused by compression. Furthermore, a multi-channel, fast volume control in the form of an automixer is common. Put simply, an automixer's task is to distribute the gains of competing signals, such as those assigned to speakers in a talk show, using normalization so that the overall gain of all signals, or the overall volume, remains constant, and the loudest signal / channel receives the largest share of the gain.Passive microphone signals are typically suppressed to such an extent during the presence of a relatively loud microphone signal from an active speaker that the proportion of ambient noise captured by these open microphones decreases noticeably. Furthermore, the automixer increases the maximum gain of a connected loudspeaker system that is possible before feedback occurs. In side-channel processing of the audio signals, also called "side-chain" processing, a control unit, also known as a compression control unit, determines a gain signal for each channel of the audio channels grouped together in an automix, for a controllable output amplifier of that channel. For this, it is only necessary that the corresponding channel levels are determined in the side-chain processing of the control unit for the individual channels that are grouped together.Additionally, the control unit's sidechain processing creates a summed level of the audio signals combined in the automix. The control signal is determined individually for each controllable output amplifier by comparing the summed level with the corresponding channel level. If the summed level in a channel exceeds the corresponding channel level, the output gain in that channel is reduced. The greater the excess, the greater the reduction. Thus, in an automixer, the individual audio signals are attenuated differently. This type of volume processing is well-suited for talk shows or similar applications, but not necessarily for other uses. Applicant: STAGETEC GmbH, September 12, 2025. Our reference: P19.090W0. Often, the same mixing console is used for talk shows, but also for mixing instrumental and vocal tracks.Consider a situation where instruments or human voices, grouped together, have already been individually pre-compressed as needed—that is, they have undergone signal processing within their respective channels, resulting in signals with controlled dynamics and requiring no further compression. If the goal is now to level the group relative to itself due to fluctuating levels among active participants, this cannot be achieved with a standard automixer. This is because weaker individual signals can be significantly suppressed by stronger signals. Manually adjusting the levels over time, however, requires considerably more effort from the user.The invention is based on the objective of creating a compression method and a compression device with enhanced capabilities for effective audio signal processing, in particular enabling automated leveling of audio signals within a group. This objective is achieved by a multi-channel compressor with the features of claim 1 and a method with the features of claim 9. Advantageous embodiments are described in the dependent claims. The invention is based on the idea of ​​creating a multi-channel compressor in which each channel has its own compressor, i.e., a controllable output amplifier, typically a voltage-controlled amplifier (VCA). A compression control device is configured to determine the individual gains, or gains, in a so-called "side-chain" processing based on the audio signals present in the individual channels.To determine control voltages. In side-channel processing of a compression control unit (CCU), the audio control channel signals (c.) are determined for the audio compression channels grouped together. m (n)) to an audio control channel sum signal (s p (n)) summed. In at least one detector unit, the individual audio control channel signals (c) are summed. m (n)) Channel level (D m (n)) and for the audio control channel sum signal (s p (n)) a sum level (S p (n)) determined. Control signals, Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO which control the individual output amplifiers and represent the gains, are determined based on the sum level and a threshold value that is adaptive. This threshold value is formed by taking a current maximum channel level (maxm{D m(n)}) is determined, to which a fixed or adjustable non-negative offset is added. This offset is not time-dependent but constant. The threshold is adaptive because the channel levels change over time. The control signals representing the gain for the individual audio compression channels of a group of channels are determined by comparing the sum level, into which their audio signals have been incorporated, with the threshold. A reduction in the gain of all channels in a group occurs when the sum level exceeds the adaptive threshold of the group. The adaptive threshold is the same for all channels in a group, i.e., a common adaptive threshold. In a preferred multi-channel compressor, the audio compression channels can be grouped into one or more groups for which compression with an automatic threshold is performed.Each audio compression channel belongs to only one group at most. In a simple embodiment, all audio compression channels of the multi-channel compressor are permanently linked to one group.In particular, a multi-channel compression method with automatic threshold setting (group compression with automatic threshold) for a plurality M of audio signals is thus created, comprising the following steps: feeding audio signals into a plurality M of audio compression channels, each of the audio compression channels having an input and an output as well as a controllable output amplifier arranged between them; optionally defining links between the individual audio compression channels to form a group of channels or several groups of channels; determining the individual gains for the output amplifiers of the individual audio compression channels of the one group or the several groups by extracting a separate gain from the audio signals of audio compression channels that are grouped together in a side-channel processing operation. Applicant: STAGETEC GmbH 12.09.2025Our reference: P19.090WOSA sum signal is generated and a sum level is determined based on the sum signal, and the control signals representing the gains for the individual audio compression channels are set by comparing the sum level, into which their audio signals have been incorporated, with a threshold value, and wherein an attenuation of the gain occurs when the sum level exceeds the threshold value, wherein the threshold value is determined adaptively for each channel group by determining a channel level for each audio compression channel of a group and, to calculate the adaptive threshold value, determining a maximum channel level of the group and increasing this by a positive offset, wherein the adaptive threshold value is used for all audio compression channels of the respective channel group. Thus, an adaptive threshold value is generated for each group of channels from the relevant channel levels.The adaptive threshold is the current maximum channel level plus a positive offset. The gain signals are derived from comparing the sum level with the adaptive threshold. Attenuation of the channels in the group begins when the sum level exceeds the adaptive threshold. If multiple groups are formed, an adaptive threshold is determined individually for each group. Defining the linkages is optional and can be omitted if all audio compression channels are permanently assigned to one or more groups, with each audio compression channel assigned to only one group.Furthermore, a multi-channel compressor with automatic threshold setting is proposed, comprising a plurality M of audio compression channels (AKK-m), each audio compression channel (AKK-m) comprising an input (Em) and an output (Am) and a controllable output amplifier (compressor VCA-m) connected between them in terms of audio signaling, and a compression control device (KSE) comprising, for each of the plurality M of audio compression channels (AKK-m), an audio control channel (ASK-m) whose audio control channel input is connected to the input of the audio compression channel (AKK-m), and an audio control channel output (StA-m) which is connected to a control input (StE) of the controllable output amplifier (VCA-m), wherein each of the audio compression channels (AKK-m) optionally has a linking device. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO is assigned to link with at least one other of the audio compression channels (AKK-m), such that several of the plurality M of audio compression channels (AKK-m) are linked or can be linked to form a group or to several groups, wherein the compression control device (KSE) additionally includes at least one summing device (SE) for forming audio control channel summation signals (see p (n)) (AVS- p) of interconnected audio control channel signals (c m (n)) the majority of the audio control channels (AKK-m) for each of the groups and at least one detector device for determining channel levels (D m (n)) for the individual audio control channel signals and sum levels (S p(n)) for the audio control channel sum signals (sp(n)), wherein the compression control device (CCD) includes a channel gain calculation device (CCD-m) which provides a target gain signal value for each audio compression channel as a function of the sum level (S p (n)), into which the audio control channel signal has been received, and calculates an adaptive reference threshold, wherein attenuation of the gain occurs when the respective sum level exceeds the respective adaptive reference threshold, wherein the compression control device (CCD) has a maximum value determination device for determining the current maximum channel level (maxm{D m(n)}) and includes a comparison threshold calculation unit that adds a fixed or adjustable non-negative offset to the current maximum channel level when generating the adaptive comparison threshold. The linking devices for the audio compression channels are optional in the sense that they can be omitted if all compression channels are permanently linked to one or more groups. To increase flexibility, a multi-channel compressor preferably includes a compression mode selection device (CMD-m) configured to set a compression mode for each of the audio compression channels (ACC-m) or group of linked audio compression channels (ACC-m). Additionally, the comparison threshold calculation unit is configured to determine the comparison threshold depending on the compression mode. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WOTo implement a further compression mode, a group compression with a fixed threshold, similar to bus signal compression without actually combining the audio signals of the individual audio channels into a bus signal before their individual compression, one embodiment provides that, as in the group compression mode with automatic threshold, the compression control unit determines the sum signal(s) for the one or more groups of interconnected audio signals or channels that are subjected to common, interrelated compression, i.e., a dynamic-limiting output gain. If a sum level, that is, a determined level of the sum signal, exceeds a threshold C definable for the respective group compression. pThis results in a reduction of the output gain of all output amplifiers of the channels coupled into a group. To achieve this, individual channels can be linked together to form one or more groups, with each channel only being able to be included in one group. Such a compressor can emulate a bus compressor, but the individual audio signals / channel signals, which are later combined into a bus signal after individual compression, are each compressed to the same degree. However, they can still be used individually as a starting point for send signals, which can then be fed to effects units, for example, to generate effect signals for further mixing. This compression mode solves problems that occur with conventional compressors that compress a signal combined into a bus signal.Often, one wants to add a separate reverb component to individual audio signals, such as a snare drum, and uses a send output for this purpose. A send output is a weighted tap of an audio signal. If the output signal of the reverb unit is mixed into the bus after passing through a group or master compressor, the ratio between the dry and reverberated signal varies because the compressor modulates the volume of the dry signal, which is included in the master bus. This case is illustrated in Fig. 1 for three channels that send their signals proportionally to the effects unit. The reverb signal, or more generally, the effect signal, is referred to as the wet signal, and the dry signal as the dry signal. This results in a varying wet / dry ratio, which is often undesirable. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO One measure against this effect, according to the prior art, is to include the reverb signal in the group sum before the compressor. This is illustrated in Fig. 2. This approach has the disadvantage that the effects unit would not be used for other sources outside the group and that the effect signal itself is compressed, which is not necessarily desirable if a particularly authentic spatial impression is to be created. The group compression mode with a fixed threshold described here solves these problems. Due to the fact that each channel has its own controllable output amplifier, a so-called automixer mentioned above can also be implemented with the same multi-channel compressor without much effort.For this to work, it is only necessary that the corresponding channel levels are determined and used in the "side-chain" processing of the compression control unit for each individual channel grouped together. Instead of a constant threshold value for all channels, which is used to emulate a bus compressor, or an adaptive threshold value, the control signal for each controllable output amplifier is determined individually by comparing the sum level with the corresponding channel level. If the sum level in a channel exceeds the corresponding channel level, the output gain in that channel is reduced. In an automixer, the individual audio signals are thus adjusted and compressed differently with regard to their dynamics. It is explicitly pointed out here that no bus signal is generated in the multi-channel compressor described here. Only in the side-channel processing, i.e.,In the compression control unit, a summed signal is generated from the audio signals that will later be combined into a bus signal, and its sum level is determined. However, this summed signal serves only for control purposes and does not represent direct signal processing of an audio signal in the sense of a bus signal. This summed signal is also referred to here as a virtual sum or virtual sum signal. It is, however, a real signal that is only a control signal or is only used for control signal processing and is not the "actual" processed audio signal. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO The created multi-channel compressor with configurable channel mappings or several predefined group mappings is thus capable of executing different compression methods, even simultaneously, for different channel groups.If the channel mapping is freely configurable, the number of group compressors and / or automixers depends solely on the group size(s) assigned to each compressor or automixer, since each channel can only be assigned to one group. If a channel is not linked to any other channel, it can be compressed individually according to a threshold that can be defined for the channel, or preferably, compression can be deactivated so that the output gain is independent of the level of the respective channel. The compression control unit (CCU) thus includes a channel gain calculation unit (CCU-M), which calculates a target gain signal (Hm(n)) for each audio compression channel as a function of the sum level S. p(n), into which the audio control channel signal has been received, and a comparison threshold is calculated, wherein the comparison threshold in group compression mode with automatic threshold is the current maximum channel level (maxm{D) increased by a fixed or adjustable non-negative offset. m (n)}) of the channels linked to a group, and if the compression mode is selectable, depending on the compression mode set by the compression mode selector (KMW-m) for the corresponding audio compression channel (AKK-m), wherein the comparison threshold differs from the channel level (Dm(n)) in automix mode and a fixed threshold constant (C) in fixed threshold group compression mode. p) is. Furthermore, a multi-channel compression method for a plurality M of audio signals is created, comprising the steps: Feeding audio signals into a plurality M of audio compression channels, each of the audio compression channels having an input and an output as well as a controllable output amplifier arranged between them; optionally defining links between the individual audio compression channels to form groups of channels; Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO Determining the individual gains for the output amplifiers of the individual audio compression channels by forming a summed signal from the audio signals of audio compression channels grouped together in a side-channel processing procedure and determining a summed level based on the summed signal,and for each audio compression channel, the control signal representing the gain is determined by comparing the sum level into which the audio signal of the audio compression channel has been fed with the threshold, with attenuation of the gain occurring when the sum level exceeds the threshold, wherein for each channel group a compression mode is selected from a set comprising a fixed-threshold group compression, an automatic-threshold group compression, and an automix.wherein, when the compression mode Automix is ​​selected for a group of linked audio compression channels, a channel level is determined for each of their audio signals, and the channel level is used as the threshold when determining the audio compression channel-specific gain; and, when group compression with fixed threshold is selected for a group, a compression constant is set and used as the threshold when determining the audio compression channel-specific gain; and wherein, when the compression mode Group compression with automatic threshold is selected for the group, the respective channel level and a current maximum of all channel levels of the group are determined for the channels of the group, and the current maximum channel level (max) increased by a fixed or adjustable non-negative offset is used as the automatic threshold. m {D m(n)}) in which the determination of the audio compression channel-specific gain is used. A preferred embodiment provides that each of the audio control channels includes a side-chain filter (SFC). These optional filters expand the possibilities for sound shaping. With regard to the above example with a multitude of drum audio signals, for example, the influence of the audio signal of a bass drum can be reduced by attenuating the bass frequencies with the aid of such a filter. These filterings influence the achievable sound. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO Another possibility for sound shaping, in particular by, for example, certain instrument tracks, i.e., certain of the audio signals, is created in an embodiment in which each of the audio control channels includes a weighting device (Wm).Such a weighting device assigns a weight to the audio signal for the corresponding audio control channel when calculating the summed signal and thus the overall level, and also influences any channel level determination that may be performed. Adding weighting allows the influence of individual signals on the group's compression to be increased or decreased. An example would be a relatively loud bass drum, whose influence would be reduced by weighting it below 0 dB. Since each signal can be weighted individually, entirely new possibilities for dynamic processing and thus sound shaping of a group emerge. To avoid distortion during compression, it is advisable to smooth the gains determined by comparing the overall level with the corresponding threshold.To avoid distortion, one embodiment provides that the signals representing the target amplifier values ​​determined in the channel gain calculation device (KVBE-m) are passed through a smoothing device before being used to drive the controllable output amplifiers. The smoothing can be implemented using a device similar to an envelope follower and can be controlled via three parameters A. s , J s , R s , be fixed, which has a response time A s (English: attack), a holding time H s (English: hold) and a cooldown period A s(English: release) represent parameters that are individually adjustable, in digital processing as multiples of sampling cycles, and each defines a smoothing phase. Preferably, the same parameters are used for all audio control channels grouped together. Other smoothing configurations are also possible. In contrast to an envelope follower, the attack phase begins when the target gain is below the smoothed gain, whereas for an envelope follower, the attack phase begins when the instantaneous level is above the smoothed level. Applicant: STAGETEC GmbH, September 12, 2025. Our reference: P19.090WO Especially in training applications where audio signals are presented as a sequence of values ​​sampled at equal time intervals in digital form, an electronic circuit in the form of an FPGA or a dedicated ASIC can be used to process the data at a significantly higher clock rate than the sampling rate of the audio signal during digitization. This allows for the implementation of circuits that, for example, perform specific calculations on an audio signal, which can then be used iteratively for the various audio control channels. Alternatively, the calculation can be performed simultaneously, i.e., in parallel, in all control channels.The processing in the compression control device is preferably carried out in stages, since digital processing in processing sections or stages reduces the complexity of the calculation and facilitates the determination of quantities, for example by allowing results or intermediate results to be used in one or more subsequent processing stages.One embodiment therefore provides that the determination of the quantities takes place in a multi-stage iterative or multi-stage parallel process, wherein in one iteration one or more of the quantities and / or one or more intermediate results are determined for one of the audio compression channels (AKK-m), wherein different quantities or different intermediate results can be determined simultaneously for different audio compression channels (AKK-M) in one iteration, wherein the iterations of a stage are executed a number corresponding to the number of multiples M of the audio compression channels (AKK-m). In a particularly preferred embodiment, the multi-channel compressor is designed to also take panning of the audio signals output from the multi-channel compressor into account during compression. During panning, the audio signal of a channel is processed with different proportions (here referred to as panning factors) Pm. 1, Pm 2, …, PmN, mixed on N output channels or panning tracks. In the stereo case, where N equals two, the panning factors are also referred to as PL(θm) and PR(θm), where an angle θm is called the panoramic angle and the panning is set individually for each channel. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO The adjustment is achieved by splitting the optionally weighted and optionally filtered audio control channel signals into N-tuples according to the panning factors in the audio control channel. For these tuple values, i.e., separately for each panning track, an N-tuple panning track sum level and an N-tuple panning track level are determined. The maximum corresponding tuple value is then determined as the sum level and the maximum channel level, i.e., the maximum panning track sum level and the maximum panning track level of the respective channel. The remaining signal processing remains unchanged.A simple design is achieved by arranging the individual audio compression channels side by side and indexing them. For each audio compression channel, it can be decided whether or not it is linked to the leftmost audio compression channel (without loss of generality). The leftmost channel cannot be linked in this way. Alternatively, it could be decided whether each audio compression channel is linked to the audio compression channel to its right. In this embodiment, the rightmost audio compression channel cannot be linked further. Linking to the leftmost audio compression channel is equivalent to linking to the audio compression channel with the next lower (or alternatively, the next higher) index for indexed audio compression channels.Especially with an analog compressor design, the linking device allows connections to an adjacent audio compression channel to be opened or closed. This causes the audio compression channels to form a group, each linked to the others and connected via lines. An audio compression channel that is not linked to its left (or right) neighbor, i.e., not connected to another channel, starts a new group from left to right (right to left). If it is not connected to the audio compression channel located to its right (or left), it forms a single-channel group. In this way, a different number of group compressors (with automatic or fixed threshold) or automixers can be configured. Each group can use a different compression mode. Applicant: STAGETEC GmbH, September 12, 2025. Our reference: P19.090WO In one embodiment, the audio compression channels (AKK-m) are indexed, and the linking device allows each audio compression channel (AKK-m) to be linked or unlinked only with an adjacent audio compression channel (AKK-m-1) with respect to indexing. All audio compression channels (AKK-m) can be linked either with an audio compression channel (AKK-m) with a lower index or all with only one audio compression channel (AKK-m) with a higher index. The fact that an audio compression channel cannot be linked further indicates that an active, further linking is not possible. However, the audio compression channel can indeed be linked to its neighbor or even further neighbors on the other side via an active linking of the other audio compression channel. For example, the first channel cannot be linked to a non-existent 0.Channels can be linked, but via an (active) link of the second channel to the first channel, and then to the second channel (and optionally to further channels). To allow audio signals to pass through the multi-channel compressor without compression, the audio compression channels preferably have an activation device, for example in the form of a switching device such as a push button, toggle switch, or similar. If the audio compression channel is not active, the output gain is not reduced. In a particularly preferred embodiment, in addition to or as an alternative to the group compression mode with automatic threshold, one of the two other compression modes described above, automix mode and / or group compression mode with fixed threshold, can be selected.Unlike the “normal” group compression mode with a fixed threshold, in the group compression mode with automatic threshold, a constant threshold is not used for determining the gain value, but rather the maximum channel level of the audio compression channels grouped together plus a positive offset ε is used as an adaptive threshold, but preferably limited to a range [T. u , T o], which is limited by a lower threshold Tu (lower limit threshold) and / or an upper threshold To (upper limit threshold). This compression mode is suitable, for example, for situations where audio signals from instruments, but also human voices, are grouped together, have been individually pre-compressed as needed, i.e., have already undergone signal processing within their channel strip, for example, of a mixing console, before being fed into the multi-channel group compressor, so that the individual signals have controlled dynamics and do not require further compression. If the aim is now to level the group relative to itself due to fluctuating participation of active participants, the user's work can be made easier if the threshold is set automatically, i.e., adaptively.This is precisely what the group compression mode with automatic threshold achieves. Assuming that, depending on the passage of a piece of music, more or fewer audio signals are involved, i.e., they have a non-negligible signal value, especially a non-zero absolute value, then no compression occurs in passages where, for example, only one instrument or singer is active, since the threshold is always above its own level. This is ensured by the positive offset ε. When multiple voices or instruments enter, compression eventually kicks in, depending on the offset ε. From a musical perspective, orienting the compression towards the loudest audio signal seems plausible. Firstly, this signal often has the highest signal-to-noise ratio; secondly, it can temporarily dominate the overall mix and is therefore advantageously used as a level reference.Assuming that the individual audio signals are mixed together via individual faders before being fed to the multi-channel compressor, and that neutral weights are applied, the dominant signal, which is louder than the rest of the audio signals in the group, would eventually mask the other signals in the same group. At that point, the compression would be reduced because the threshold value relative to the overall level has automatically increased when the other audio signals no longer contribute significantly. However, if the group signals are all approximately the same volume, then the level of virtually every currently active signal, including the level of the loudest signal in the group, provides a good reference point for setting the compression gain.In such a case, the group compressor behaves approximately the same regardless of the absolute level of the individual signals, and its level reduction increases with the number of similarly loud signals involved. This allows the overall group to be balanced in a controlled manner, as mentioned. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO If there is also a positive weighting of the signal, for example, a solo or lead singer, the reference to this signal can be enforced. Vocal doubling then triggers group compression depending on the level. The upper limit, To, can optionally be used to prevent the overall group volume from becoming too high. The lower limit, Tu, optionally ensures that no compression occurs in quiet passages with a low overall group volume, so that the group is not masked compared to other signals in the mix.Level determination can be performed in various ways. A unit used to determine a level is also called a detector, and the process of determining the level is called detection. A fundamental distinction is made between level determination based on an amplitude value or the signal power. In the first case, the absolute values ​​of the audio signal are used, each plus a very small positive offset (peak detection). In the second case, the squares of the absolute values ​​of the audio signals (power data), optionally increased by another small offset, are used. There are different variations for the different detection methods, as explained in more detail below. Preferably, the determined values ​​are logarithmized with respect to an arbitrary base. These logarithmic values ​​are then referred to as levels. The gain values ​​are then also calculated using these logarithmic values.Different methods can be used here, referred to as the soft-knee and hard-knee variants. In the hard-knee variant, the compression, i.e., the reduction of the target gain, begins abruptly. The function curve exhibits a kink at this point. In logarithmic space, the two sections are linear. In the soft-knee variant, this kink region is replaced by a segment of a polynomial, where the function values ​​of the polynomial function and the first derivatives of the polynomial at the boundaries of the replacement region coincide with the function values ​​or derivatives of the two straight segments of the gain function, each of which is a function of the difference between the sum level and the corresponding threshold. In the hard-knee variant, the target gain, i.e., the logarithmic gain Gm(n), is equal to... Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO an optional negative impact factor I multiplied by the maximum of the difference between the sum signal level Sp(n) and the threshold Tm on the one hand, and zero on the other. This means that the logarithmic target gain Gm(n) is zero as long as the sum signal level Sp(n) is less than the threshold, and otherwise the difference between the sum signal and the threshold, possibly multiplied by the optional negative impact factor. The impact factor I, also called the effect factor or influence parameter, lies between 0 and 1. If the logarithmic gain Gm(n) is zero, there is no attenuation of the output gain. The larger the absolute value of the logarithmic gain value, which is always negative, the stronger the compression. The logarithmic gain value is converted back into a linear target gain value by exponentiation before any optional smoothing of the gain.The invention is explained in more detail below with reference to the drawings. These show: Fig. 1 a prior art signal processing system in which a bus signal formed from three signals is compressed and reverb effects are subsequently added to it; Fig. 2 a prior art signal processing system in which reverb effects are added to a bus signal formed from three signals before compression; Fig. 3 a schematic representation of a multifunctional multi-channel compressor; Fig. 4 a signal flow diagram of a detector for single dynamic processing of a group of signals; Fig. 5 a schematic representation of an exemplary configuration of a multifunctional multi-channel compressor with two group mixes, an automix, and single compression; Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO Fig. 6 a schematic representation of embedding the automixer / compressor in a flexible signal processing path; Fig.Fig. 7 A schematic representation of a sequential processing architecture and a timing scheme for processing M channels; Fig. 8 A schematic organization of a memory for signal processing, corresponding to the configuration according to Fig. 5; Fig. 9 A schematic representation of a control panel of a console system with group processing unit; Fig. 10 A schematic representation of an automixer or group compressor as a chain of a single basic element (U1, U2, …); Fig. 11 Principle diagram for a basic element for dynamics processing according to variant 1; Fig. 12 Principle diagram for a basic element for dynamics processing according to variant 2; Fig. 13 A schematic representation to illustrate its use: a multifunctional multi-channel compressor for stereo processing of mono and / or stereo channels distributed in the panorama.Figure 1 shows a signal processing device 1 according to the prior art, in which signals from three channels 11, 12, 13 are combined in a summing amplifier 20 to form a bus signal 30. This bus signal 30 is fed to a group compressor or bus compressor 100, which generates a compressed bus signal 110. A so-called send channel 41, 42, 43 is branched off from each of the three channels 11, 12, 13. Each of the send channels 41, 42, 43 includes a weighting device 51, 52, 53, which is designed, for example, as an adjustable amplifier. The signals of the three send channels 41, 42, 43 are combined in a further summing amplifier 60 to form a summed signal 70, which is fed to an effects unit 80. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090 The effect signal 90 output by the effect device 80, for example a reverb-generating device, is combined with the compressed bus signal 110 in yet another summing mixer 120 to form the output signal 130.Channels 11, 12, and 13, for example, represent audio tracks of a miked drum kit, consisting of the kick, snare, and hi-hat tracks, which are summed to form a group signal, also called bus signal 30, i.e., a separate unit or "group." This typically involves routing a specific selection of tracks to a mono, stereo, or surround bus that represents the group. Such a bus signal 30 can then be subjected to subsequent processing steps, such as, in the example shown, the combined compression of the bus signal 30 representing a drum track to reduce the dynamic range. This often results in a subjectively perceived compression or tightening of the signal, which is a desired effect.In this prior art processing method, a problem arises: applying desired effects to individual or all signals—for example, adding a separate reverb component to a single signal like the snare drum, or adding a reverb component to all signals—leads to undesirable effects. If send channels 41, 42, 43, or send outputs are used proportionally for this purpose, the ratio between the dry and reverberated signal varies during signal processing according to Fig. 1, since the compressor modulates the volume of the dry signal, which is included in the overall signal mix. This varying wet / dry ratio is often undesirable. Fig. 2 shows another possible prior art signal processing method to circumvent this disadvantage. In contrast to the embodiment shown in Fig.In Figure 1, the effect signal 90 is combined with the bus signal 30 before group or bus compression in the bus compressor 100 and then in the further summing compressor 120, which is arranged before the bus compressor 100 and generates the output signal 130. Identical technical features are indicated with the same reference symbols in all figures. This variant has the disadvantage that the effect unit 80 could not be used for other sources outside the group and that the effect signal 90 itself is compressed, which is not necessarily desirable if a particularly authentic spatial impression is to be created. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO Both signal processing approaches are not entirely satisfactory depending on the requirements. Figure 3 schematically shows a proposed multifunctional multi-channel compressor 200 and a signal processing method that can be implemented with it.The problems inherent in this technology can be circumvented by subjecting the individual tracks 211, 212, and 215 to a common dynamic processing, but without first combining them into a group signal. This allows the individual signals to remain available after dynamic processing, i.e., compression, and to be accessed via send channels. The multifunctional multi-channel compressor 200 comprises M audio compression channels 210, 210-1, 210-2, ..., 210-M. Each of the audio compression channels 210, 210-1, 210-2, ..., 210-M includes an input 220, 220-1, 220-2, ..., 220-M. Up to M signals are fed into the multi-channel compressor 200, originating, for example, from M channel strips 180, 180-1, 180-2, 180-M of a mixing console. These channel strips 180 are indicated by the output faders 190, 190-1, 190-2, ... 190-M.Each of the audio compression channels 210, 210-1, 210-2, …, 210-M further comprises an output 250, 250-1, 250-2, …, 250-M, as well as a controllable output amplifier 240, 240-1, 240-2, …, 240-M arranged between them. This is, for example, designed by means of a voltage-controlled amplifier (VCA). The controllable output amplifiers 240, 240-1, 240-2, …, 240-M are also referred to as compressors. The multi-channel compressor 200 comprises a compression control device 300, which includes an audio control channel 310, 310-1, 310-2, …, 310-M for each audio compression channel 210. An audio control channel 310 comprises an input 320, which is connected to the corresponding input 220 of the corresponding audio compression channel 210. Each audio control channel 310 further comprises, in the illustrated embodiment, a side channel filter 330, 330-1, 330-2, …, 330-M and a weighting device 340, 340-1, 340-2, …, 340-M.A channel gain calculation device 350 determines a target gain signal 390 as a control signal for each audio compression channel 210, 210-1, 210-2, …, 210-M or its controllable output amplifier 240, 210-1, 240-2, …, 240-M. The compression control device 300 summes all optionally filtered and optionally weighted audio control channel signals 360, 360-1, 360-2, …, 360-M of the audio control channels 310, 310-1, 310-2, …, 310-M in a summing amplifier 355 to form a summed control signal 370. The compression control device 300 further comprises at least one detector device 380 which, starting from the sum control signal 370 and the optionally filtered and optionally additionally or alternatively weighted audio control channel signals 360, 360-1, 360-2, …, 360-M, generates a preferably logarithmic sum level 375.Depending on the compression mode, the detection device 380 also determines a preferably logarithmic channel level 365, 365-1, 365-2, …, 365-M for each audio control channel signal. The sum level 375 and, if applicable, the channel levels are fed to the channel gain calculation unit 350, which generates a target gain signal 390 in a group compression mode. The value of this target gain signal indicates the gain for the corresponding audio compression channel 210, 210-1, 210-2, ..., 210-M. In principle, various compression modes are possible, so other embodiments may include a compression mode selection device (not shown). The individual target gain signals 390-M control the compression by means of the controllable output amplifiers 240 in the individual audio compression channels 210. The embodiment of the multi-channel compressor in Fig. 3 can be implemented using analog or digital technology.The diagram is to be understood as a signal flow diagram only and has no direct implications for the implementation. M input channels pass through their channel strips 180 (optional, not shown) and corresponding output faders 190 before entering the multi-channel compressor 200. In the summing unit 355 of the compression control unit 300, a weighted "virtual sum", the sum control signal 370, is formed from optionally pre-filtered channels or audio control channel signals 360, which is fed to the detection unit 380. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO The compression control unit 300 controls M parallel digital or analog output amplifiers 240 (voltage-controlled amplifiers = VCAs) via corresponding control signals, the target gain signals 390. As with all audio compressors, the dynamic range is limited by reducing the output gain as soon as a signal, in this case a group compression, exceeds the virtual sum, i.e.The summing control signal 370 exceeds a threshold, possibly gradually. Since the individual signals, i.e., the output signals 260, 260-1, 260-1, …, 260-M of the audio compression channels 210, 210-1, 210-2, …, 210-M, remain as such, they can be used for send mixes 510, 510-1, 510-2, …, 510-M. In the example shown, individual portions, weighted via optional weighting devices 520, 520-1, 520-2, …, 520-M of the channels, are sent to an effects unit 550, whose output signal 560, together with a bus signal 270, is summed in a master bus summer 290 to form a master bus signal 295. The bus signal 270 is first summed from the output signals 260, 260-1, 260-1, …, 260-M of the channels in a bus summing unit 280. The addition of weighting allows the influence of individual signals on the compression of the group to be increased or decreased.An example would be a relatively loud bass drum, whose influence is reduced by weighting it below 0 dB. Since each signal can be weighted individually, entirely new possibilities for dynamic processing and thus sound shaping of a group arise. Weighting represents an alternative or supplement to the use of the usual so-called sidechain filters. These are also included in the arrangement as "SCF" blocks 320 (optional), which further expands the sound shaping possibilities. With regard to the bass drum example, its influence can also be achieved by attenuating the bass frequencies using a filter, leading to slightly different sonic results. In addition to the summed signal 370, the pre-weighted and filtered audio control channel signals 360 and 360-m, as shown, also go directly to the detection unit 380. This unit determines a channel level.They are required for an automix mode and a novel, self-inventing special compression mode with automatic threshold, in which the threshold is determined automatically. This method is explained in more detail below. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO The terms "virtual bus sum" and "virtual bus compressor" refer to the fact that such a "bus sum" and such a "bus" only exist in the sidechain path and ensure that the dynamic processing can behave in the same way as with a summation according to Fig. 1. The method is therefore not to be confused with virtual, software-modeled audio effects. Multi-channel bus compressor as a modification of an automixer It can be seen that the signal flow shown in Fig. 3 essentially corresponds to that of an automixer, but performs compression as described.An automixer is a well-known device whose function, in short, is to distribute the gains of competing signals, such as speakers in a talk show, using normalization so that the overall gain of all signals, or the overall volume, remains constant, and the loudest channel receives the largest share of the gain. Passive microphone signals are typically suppressed during the presence of a relatively loud microphone signal from an active speaker, significantly reducing the amount of ambient noise picked up by these open microphones. Furthermore, the automixer increases the maximum gain of a connected loudspeaker system before feedback occurs. In addition to the weighted sum, the automixer also requires the individual channel signals for its detector. Bass or low-frequency suppression is also possible.Low mid-frequencies are reduced by pre-filtering, which also reduces intermodulation distortion caused by the control process with its finite response times. In contrast to the previous consideration regarding high-energy, loud signals (e.g., bass drum), the filters thus fulfill a function that weighting alone does not. If a digital signal processor is implemented using FPGA technology or even as a custom integrated circuit (ASIC) and includes an automixer, then chip area is typically allocated to this processing unit. The multifunctional multi-channel compressor has the advantage that, by modifying its functionality, it can implement both a multi-channel compressor and an automixer without significant additional effort. The functionality of the automixer and the multi-channel compressor is therefore defined mathematically in the following section.Mathematical description of the method of a multi-channel virtual bus compressor, either on its own or as an automixer. The following definitions are given for discrete-time digital systems, which, due to technological advantages, also represent the most common application today. An analog implementation will be presented later. When "time-invariant" coefficients or constants are mentioned below, this means that no time dependencies exist as long as these coefficients are not changed directly or indirectly by the user. Assume a discrete-time sampling system with a constant sampling rate fT = 1 / T, where T denotes the sampling period. For each sampled value x(nT) of a signal in the system at time nT, an integer time index n suffices. Therefore, x(n) is simply written instead of x(nT). Equations are generally abbreviated as "Eq." and figures as "Fig."Equations are numbered consecutively using integers. For related equations, these are indicated by a common integer and a sub-numbering in the form of an incremental decimal place. Alternative equations that replace the corresponding originals are marked with a lowercase letter suffix, beginning with "b". In some cases, the alternatives receive their own numbers instead of a suffix, which is then explicitly indicated. Given M synchronous input signals, i.e., those subject to a sampling rate, a. m(n), m=1…M of the arrangement to be defined in Fig. 3, where m denotes the channel index and n the time index. These input signals are usually post-fader output signals from channel strips on a mixing console. In the first version of the description, for the sake of simplicity, all M channels are assumed to be involved in the processing. As will be shown, unused channels can still be excluded from processing and pass through the arrangement without modification. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO The signal flow diagram above in Fig. 4 is a detailed view of a compression control unit 300 according to Fig. 3. Based on the following equations, the letters mark the signals of the same name after the individual process steps have been completed. Reference symbols are the same as those in Fig. 3. Preprocessing As shown, the input signals a m (n) initially pre-filtered optionally. For the output signal b m(n) of the m-th channel after passing through a linear, time-invariant, feedback filter of K-th order with the coefficients B k and A k As is well known: The signals are then weighted by applying non-negative factors w mto be multiplied.^^(^) = ^^(^) ∙ ^^ (2) A neutral weighting corresponds to a weight of w=1. By setting its weight to zero, the influence of the m-th channel on the compression or the automix can also be completely neutralized. Note that the process order filtering-weighting can also be reversed and leads to the same result. For both, the automix mode and the group compression modes, the sum of the preprocessed audio control channels 360 is required:^(^) = ∑^ ^^^^ ^ (^) (3) Detection The individual channels and the sum now each pass through a detector element 341, 341-1, 341-2, …, 341-M, 341-S with the aim of estimating either the peak value or the average power. In the second case, the effective value will be determined later. Applicant: STAGETEC GmbH, September 12, 2025. Our reference: P19.The response times of the summation detector and channel detectors, derived from coefficients, are chosen to be the same or similar, with values ​​suitable for practical applications. Two heuristic variants (6.1) and (6.2) for peak value estimation and two further variants (7.1) and (7.2) for average power are introduced below. Each of these detectors is, in principle, an operator with memory in the form of a number of state variables, which provides an output value y(n) for a given input value x(n). First, define ^^(^) = |^(^)| + ^^ (4). xa(n) here corresponds to the absolute value of x(n) plus an optional, very small offset o1, which is introduced for numerical reasons. xp(n) is the power square, where a small, optional offset o2 can also be used. The first peak detector type determines the maximum of the last L rectified samples. ^^(^) = max {^^(^), ^^(^ − 1), … , ^^(^ − ^ + 1)} (6.1) The second peak detector type estimates the peak value using an envelope follower with a response time, hold time, and decay time, represented in that order by three coefficients Ae, He, Re, where He is an integer number of samples. The state of the detector consists of the previous output value y(n-1) and a counter value z(n-1). The current initial value and the new meter reading are calculated as described below. Three cases must be distinguished. (6.2) Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO ^(^) ≔ ^(^ − 1) = 0 for ^^(^) ≤ ^(^ − 1) and ^(^ − 1) = 0. During the attack phase, when the input value exceeds the detector state, the counter is set to the hold time. Once the input signal falls below the current estimate of the peak value after the attack phase, the hold phase begins, and only after the counter has completely counted down to zero does the release phase begin, unless the attack phase is triggered again. The attack time and hold time can also be set to zero. Note that y(n) cannot take on negative values, since the input current is always positive. The first type of power detector consists of the mean of the squares of the last L samples: The second type of power detector uses a feedback first-order filter with coefficient c0, where |c0|<1:^(^) = ^^ ∙ ^^^(^) − ^(^ − 1)^ + ^(^ − 1) (7.2) For the application, choose one of these four detectors and apply this type to the sum and the individual channels, where the coefficients used can vary slightly between the sum detector and the channel detectors. Let Dch{x(n)) denote the operator for the individual channels c m(n), which results from equations (4) and (5) and the detector selected from the palette (6.1), (6.2), (7.1), (7.2). Similarly, denote by Ds{x(n)} the operator according to the selected detector for the sum s(n). These memory operators operate not only on the current sample, but also on past samples, i.e., the entire sequence. Then, for the individual signals in Eq. (2), a total of M detector signals d result. m(n) and for the sum in Eq. (3) a summing detector signal se(n):^^(^) = ^^^{^^(^)} (8)^^(^) = ^^{^(^)} (9) When using the automix mode combined with one of the presented power detectors, the summing detector signal should be limited from below so that the sum does not become smaller than individual signals in the case of out-of-phase, correlated recordings. The limit is defined as the sum of the power contributions of the individual channels, which is formed at a summing amplifier 342: ^^(^) = ∑ ^ ^ ^^ ^ ^(^) (10) This limit corresponds approximately to the power sum of M uncorrelated signals. Then the sum detector signal is to be bounded from below and the following is to apply instead of Eq. (9): ^^(^) = max {^^{^(^)}, ^^(^)} (11) max{a,b} denotes here and in further equations the maximum value of a and b. In general, both in the group compression modes and in the automix mode, the task of the entire detector circuit 380 together with the channel gain calculation device 350 is to calculate for each channel m a continuous, linear, “raw” target amplitude gain H m (n) to be calculated, which is then smoothed. Without considering further factors that will be introduced, the following would apply when using peak value detectors, specifically for the automix mode: Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO However, when using power detectors, the following would have to apply: As can be seen, the gain in automix mode is limited to a value of 1 (corresponding to 0 dB). In addition, there is the compression mode, which requires a different rule. Equations (12) and (13) above merely specify how an automixer should behave, but are not used in this form in connection with the procedure. The calculation of the target gain H can instead be simplified and performed in the logarithmic domain. Any base β is suitable for this purpose, and without loss of generality, the base β=2 is chosen here. A different base only changes the scaling. The detection device includes logarithmic elements 343-1, 343-2, ..., 343-M, 343-S. The detector signals for channels and sum are given by the following expressions. When using peak values, which can also be called peak mode, the logarithmic detector signals D can be calculated independently of the mode (automix or compression). m(n) and S(n) write the following: ^^(^) = ^^^^(^^(^)) (14.1)^(^) = ^^^^(^^(^)) (14.2) When using power detectors, which can also be called RMS mode, effective values ​​are obtained by halving the logarithmic detector values, so that both detector types can subsequently be treated in the same way. Thus, the detectors operate as RMS detectors (RMS = root mean square). Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO ^^(^) = ^^^^(^^(^)) / 2 (15.1)^(^) = ^^^^(^^(^)) / 2 (15.2) Note: D m(n) in Eq. (14.1) or (15.1) shall be called "channel level". S(n) in Eq. (14.2) or (15.2) shall be called "sum level". Target Gain: Since the levels of the audio control channel signals and the sum of the audio control channel signals are in the logarithmic range, a logarithmic target gain G can be determined in the channel gain calculation device 350 with determining elements 351-m. This is transformed back into the linear range into the amplitude gain H by inverse transformation in power converters 352-m before smoothing is performed in smoothing elements 354-m. At this point in the determination of the logarithmic target gain G, the method distinguishes between two different target curves, depending on whether the target gain as a function includes a "hard" or "soft" transition.Furthermore, the three operating modes are distinguished here: automix, group compression with a fixed threshold, and group compression with an automatic (adaptive) threshold. "Hard-Knee" Variant: For the so-called "hard-knee" case, a target gain G is defined in the logarithmic range relative to the selected base β for each channel. m (n) (“gain”), a generalized threshold T m (n) (“threshold”) and an influence I m(“impact”) with the following relationship: ^^(^) = −^^ ∙ max{^(^) − ^^(^), 0} = −^^ ∙ F(^(^) − ^^(^)) (16) with Applicant: STAGETEC GmbH 12.09.2025 Our Ref: P19.090WO^(^) = max {0, ^} (17) Due to the limitation in (16) using the “kink function” F(x), the target gain G can only take on negative values ​​or the value 0. As long as the sum S(n) remains below the threshold value, no gain reduction takes place (i.e., G=0). For the influence parameter I, 0 ≤ I ≤ 1 always holds. “Soft-knee” variant: Instead of approach (16), softer, more transparent curve transitions of the target gain can be achieved by using a polynomial P. s (x) is used, which replaces the kink function F(x) in the right-hand side of (16) within a chosen logarithmic transition region W = [-δ / 2, δ / 2] and operates on the difference x=ST. In this variant, the logarithmic gain is defined as ^^(^) ≔ −^^ ∙ P ^^(^(^) − ^ (^)) for −^ / 2 ≤ ^(^) − ^^(^) ≤ ^ / 2 (18.1) or for ^(^) − ^^(^) < −^ / 2. Only within the region W, where (18.1) holds, does the curve in (18) deviate from Eq. (16). The polynomial Ps(x) used has at least second order and satisfies the property that the function value Ps(x) and the first derivative Ps'(x) at the transitions x=-δ / 2 and x=+δ / 2 coincide with the kink function F(x) or its derivative, respectively: ^^′(−^ / 2) = 0 (19.3)^^′(+^ / 2) = 1 (19.4) Operating Modes Applicant: STAGETEC GmbH 12.09.2025 Our Ref: P19.090W Now we can turn to the different operating modes. Another quantity is introduced, the maximum level within the channels at sample time n. The maximization is performed via the channel index m on a maximum finder 345.^^^^(^) = ^^^^{^^(^)} (20) Another quantity is the following adaptive threshold, which results from the maximum level plus a non-negative offset ε and, optionally, an upper limit T. oand / or optionally lower barrier T u shall be subject to:^^(^) = ^^^{^^^{^^^^(^) + ^, ^^}, ^^} (21) Depending on the operating mode, the channel-related threshold T is now m In Eq. (16) and Eq. (18), different quantities are assigned according to the following table: Table 1: Threshold for channel m depending on the operating mode Operating mode Automix Group compression Group compression with fixed threshold automatic threshold Threshold Channel level Threshold constant Adaptive threshold T m D m C Ta 1 . AutomixIn "Automix" mode, according to Table 1: ^^(^) ≡ ^^(^) (22) According to Eq. (16) or (18) and (22), the sum level of each audio control channel is compared against its own audio control channel level. For the hard-knee variant (16), this corresponds in the linear range to the quotient in Eq. (12) for peak detection or Eq. (13) for RMS detection. However, the method is extended such that the logarithmic gain G can be adjusted, i.e., attenuated, using the impact parameter I, and that the soft-knee approach in Eq. (18) is available as an alternative. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO From the user's perspective, the louder the individual channel level is in relation to the sum level, the less level reduction takes place. With only one signal in the sum, apparently no level reduction occurs.With a dominant signal in the form of an active speaker, weak signals are strongly suppressed, while the dominant signal itself is hardly affected, depending on the noise component. Pathological cases, where strongly correlated but antiphase signals are present in different channels and cause the sum level S(n) to be potentially lower than one of the channel levels involved, are handled by limiting the gain in Eq. (17). 2. Fixed-Threshold Group Compression In "fixed-threshold group compression," sometimes also referred to simply as "group compression" or "simple group compression," the varying channel level, acting as the threshold, is replaced by a threshold constant C.It holds true that: ^^(^) ≡ ^ (22b) As can be shown by calculations, this results in dynamic processing that corresponds to that of a forward-facing compressor which uses the selected detector for the weighted channel sum, i.e., operates in peak or RMS mode. With respect to the usual compressor ratio R, the following simple dual relationship with the impact factor I can be derived for the hard-knee case: ^ = 1 / (1 − ^) (23.1)^ = 1 − 1 / ^ (23.2) When the impact factor approaches the value 1, the ratio R approaches infinity, resulting in a limiter. One now generally has the choice between a global impact parameter I, which is the same for all channels, and individual impact parameters Im, an impact parameter I. mper channel. The latter case allows the influence of the group-related compression to be set individually for each channel, giving the channels an individual ratio. This can be used, for example, to ensure that a lead vocal in an overall mix is ​​hardly reduced in level under strong group compression, even though its contribution to the overall level S(n) may be significant. Note that S(n) for automixers and multi-channel compressors is a mono sum. A stereo version for the multi-channel virtual bus compressor is described below. The mono sum determines the control behavior, even if the channels are subsequently panned and summed. 3.Group compression with automatic threshold (adaptive threshold) In this case: ^^(^) ≡ ^^(^) (22c) This new mode differs from simple fixed-threshold group compression in that the adaptive threshold Ta is used instead of the constant C. According to Eq. (21), Ta is formed for each time n from the instantaneous maximum channel level plus an offset ε, preferably restricted to the range [Tu,To]. The relationship between impact and ratio is the same as in simple group compression, i.e., Eq. (23) holds. The basic idea of ​​this mode of operation has already been briefly explained above.It is assumed that instruments or human voices grouped together have already been individually pre-compressed as needed, meaning they have undergone signal processing within their respective channels. Therefore, these signals, taken individually, have controlled dynamics and do not require further compression. If the goal is to level a group relative to itself due to fluctuating levels of active participants, the user's task can be simplified by setting the threshold automatically, i.e., adaptively. Applicant: STAGETEC GmbH, September 12, 2025. Our reference: P19.090WO. It is assumed that, depending on the passage of a piece of music, more or fewer signals are involved. With a single signal, compression apparently does not occur, as the threshold is always above its own level. When additional voices are added, compression eventually kicks in, depending on the offset ε.Orienting oneself towards the loudest signal according to (21) seems musically plausible. Firstly, this signal often has the highest signal-to-noise ratio; secondly, it can temporarily dominate the overall mix and is therefore advantageously used as a level reference. Let us assume for a moment that the tracks are indeed mixed together according to the fader positions and that neutral weights are present. If the dominant signal is significantly louder than the rest, other signals in the same group would eventually be masked. In that case, however, the compression is reduced because the threshold has automatically increased relative to the sum level if the other signals do not contribute significantly to it. If, on the other hand, the group signals are all approximately the same volume, then the channel level of virtually every temporarily active signal, and thus also the channel level of the currently loudest signal in the group, provides a good reference value.In this case, the compressor behaves approximately the same regardless of the absolute channel level of the individual signals, and its level reduction becomes more pronounced the more (similarly loud) signals are involved. This allows the overall group to be balanced in a controlled manner, as mentioned. By positively weighting one signal, for example, a lead singer's, the reference signal can be enforced. Vocal doubling then triggers group compression depending on its level. The upper limit T. o This can optionally be used to prevent the overall volume of the group from becoming too high. The lower limit T uThis, in turn, optionally ensures that no compression occurs in quiet passages with low overall group volume, so that the group is not masked compared to other signals in the mix. Potentiation and smoothing of the applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO The resulting logarithmic "raw" target gain G m (n) for the m-th channel is transformed by exponentiation to the chosen basis β=2 to obtain the linear gain H m (n) to obtain. As already mentioned, choosing other bases yields equivalent results. As with any conventional compressor, smoothing the target gain with asymmetric filtering is an essential part of the functionality to avoid significant distortion. The smoothing works similarly to the threshold detector in (6.2), i.e., with an attack time, a hold time, and a release time, represented in that order by three coefficients As, Hs, Rs, where Hs is an integer number of samples. However, unlike the detector, the attack phase here begins when the linear target gain H is less than the smoothed gain J. Each smoothing element, in turn, has two states: the previous output value J and the current output value J. m (n- 1) and the counter reading Z m (n-1). The cooldown is usually chosen to be significantly longer than the attack time. Attack: (25)^^(^) ≔ ^ ^ ^^ ∙ ^^ (^) − ^ (^ − 1)^ + ^^(^ − 1)^^(^) ≔ ^ ^^ for ^ (^) < ^^(^ − 1) Hold:^^(^) ≔ ^^(^ − 1) ^^(^) ≔ ^^(^ − 1) = 0 for ^^(^) ≥ ^^(^ − 1) and ^^(^ − 1) = 0 Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO The described variant operates in the order of exponentiation-smoothing. For compressor mode, a reverse order is also possible, i.e., first smoothing in the logarithmic range and then exponentiation. For automix mode, set As=Rs and Hs=0. This results in a first-order low-pass filter with feedback, which ensures that the smoothed gain does not exhibit abrupt jumps. Output Signals The output signal ym(n) for the m-th channel, i.e., the audio compression channel, is the product of the input signal, smoothed gain, and an optional makeup gain V ≥ 1 for compressor operation. In this way, all M output signals are obtained. ^^(^) = ^^(^) ∙ ^^(^) ∙ ^ (26) The input signal a m could also be in opposition to the reinforcement J mhave a time lead or lag in integer multiples of samples. This would not substantially change the method. In Eq. (26), time differences are not taken into account. Organization of single and multiple automixes and group compressions. In the method presented so far, all M channels are involved to obtain a single automix or a multi-channel, i.e., group, compression. The term "group processing" can be used here as a general term independent of the operating mode. A channel can be removed from the group processing to which it is assigned by setting its weight and its impact factor to zero. Several group processing operations can exist side by side. In this case, the signal flow in Fig. 4 can simply be imagined as vertically multiplied, so that further inputs and outputs are added. This is legitimate insofar as an applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.The 090WO input channel is assigned to at most one group processing operation. Each duplication has its own channel sum. The numbering of the input channels is variable. The channels assigned to a group processing operation do not necessarily have to be adjacent on the mixing console. Flexible, sequential implementation of the method: There are many possible implementations. The signal flow diagrams in Fig. 3 and Fig. 4 are to be understood as exemplary embodiments and define only one exemplary implementation. A slightly modified variant, which can be implemented particularly elegantly, is described below. As before, let M channels be given as inputs. This time, however, it should be possible to process channel bundles separately. An example configuration for this approach is shown in Fig. 5 for M = 12 channels.Audio compression channels 210-1 to 210-3 are subjected to a group compression 710-1, and audio compression channels 210-4 to 210-7 are subjected to a different group compression 710-2. Audio compression channel 210-8 is not processed at all, while audio compression channels 210-9 to 210-11 are components of an automixer 720. Audio compression channel 210-12 undergoes a standard (group) compression 710-3, with the group comprising only this single audio compression channel 210-12. Each group processing and individual processing operation should have its own parameter sets. They are therefore distinct from one another. Although they have separate signal processing, they can also include the same parameters in their own parameter sets.The limitation of this approach, intended to enable ease of use and / or simplify the calculation of individual signals, is that the channels belonging to a processing operation must be arranged adjacently on a user interface or be addressable with adjacent indices. An input routing matrix can be provided in the system, which, by reordering, ensures that physical inputs are fed into the desired adjacent channel strips. Such a system is indicated by way of example in Fig. 6, where the optional channel strips in the middle are not specified in detail. Incoming signals 810-1,… 810-M are reordered in a router 820 so that the signals to be compressed together are fed into adjacent channel strips, for example, of a mixing console 830, and then fed to a multi-channel compressor 200.This implementation uses a sequential processor which generates a vector of M output channel samples from a vector of M input channel samples with time index n within a sampling period. This processing is shown by way of example in Fig. 7. The delivery of the channels can be carried out, for example, using a time-slot method, where, as shown, the sample data arrives sequentially on a data bus at short intervals, is processed, and output sequentially in batches. While the input data 910 arrives, output data 920 can be output. During the time slot 900(n), the processing 950 of a batch of sample data takes place. This principle is efficient, but not mandatory. A processor 930 has an arbitrarily configured internal or external memory 940 for storing intermediate results, states, and the like.To implement flexible channel partitioning for different group processing, partial sums must be calculated instead of a global sum s(n) according to Eq. (3). The same applies to the sum of the power shares su(n) according to Eq. (10) if an RMS detector is used in automix mode. Likewise, determining a maximum according to Eq. (20) only extends to those audio compression channels that are included in the maximum calculation, i.e., to the audio compression channels of a group that require this maximum calculated from the set of its channels as its size. Simple two-step procedure for determining partial sums and partial maxima. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO. Partial sums are defined in this context as sums of channels of a group. The following approach can be used to determine such partial sums: A binary combination vector v. mis defined such that v m = 1 should hold if and only if channel m is "coupled" to the previous channel m-1, i.e., this and the previous channel both belong to the same group processing, and v m = 0 should then apply if these channels belong to different processing operations, i.e., groups. With this definition, v1 for the first channel m=1 is always equal to 0, since it has no previous neighbor. The following pseudocode can be executed on a system to compute arbitrary non-overlapping partial sums, so that ultimately an example configuration as in Fig. 5 can be implemented. Pseudocode 1: Summation and distribution of the partial sums. Iteration 1: For all indices m from 1 to M, execute (loop) if v m = 0 (new sum, since there is no predecessor) set a := x m (first addend) set i := m (remember base index i) set Ki := a (store intermediate result in cell i) else a := a + x m(accumulate a with the next component) set Ki := a (store intermediate result in cell i) End of case distinction End of loop Iteration 2: For all indices m from 1 to M execute (loop) if v m = 0 (new sum is due, since there is no predecessor) set a := K m (Retrieve stored complete sum) End of condition set ym := a (store sum in output vector under index m) Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO End of loop In the code above, x m An input vector, which in turn can be a memory area. Then there are the temporary variables or registers a (data value or accumulator), m (running index), and i (base index). K mis a vector with M elements of some memory area, whose elements are addressed using index m. Step 1 is used to determine the (partial) sum of each group, while step 2 serves to distribute the sums to all relevant elements of the associated groups. After both steps have been completed, all partial sums have been determined. Note: In iteration 1, data is temporarily stored after each step (cell i). This is not strictly necessary, as only the last summation of a partial sum needs to be stored. Partial maxima of an input vector can also be determined in a similar manner: Pseudo-code 2: Determining maxima of individual groups Iteration 1: For all indices m from 1 to M, execute (loop) if v m = 0 (no predecessor → new maximum determination) set a := x m(first value) set i := m (remember base index) set Ki := a (store intermediate result in cell i) else if a < x m (new, temporary maximum found) a := x m (Note new maximum) set Ki := a (store this intermediate result in cell i) End of case distinction Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO End of case distinction End of loop Iteration 2: For all indices m from 1 to M execute (loop) if v m = 0 (new group is coming up) set a := K m(fetch stored maximum) End of condition set ym := a (store current maximum in element m) End of loop Step 1 is used to determine the (partial) maximum of each group, while step 2 serves to distribute each maximum to all relevant elements of the corresponding group. Inclusion of partial sums and maxima in the procedure The following steps are now carried out to implement a partitioning of the channels for different group processing. First, the intermediate results c m (n) are determined in equation (2) and are stored as data in memory. As a reminder, this data represents the current signal values ​​after filtering and weighting. To extend the originally outlined procedure, the sum s(n), defined in equation (3), is now replaced by a vector of M sum values ​​s. m (n) replaced. The component s m(n) here represents a sum applied exclusively to channel m. To determine the partial sums, the two-step procedure in Pseudo Code 1 is used above, where cm(n) takes the role of the input vector with elements x. m slips, i.e. x m (n) := c m (n), and the output vector y m the sums s m (n) results in, therefore s m (n) := y m (n). The retention of the time index indicates that this two-step process is repeated with each sample n. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO It is noted that only a single group processing is performed for all audio compression channels, all s m(n) are equal and s(n) correspond as before. In the case of multiple group processing operations, the sums of the elements of a common group are identical. Similarly, the lower bound su(n) from Eq. (10), which is only needed for the automix mode in combination with RMS detectors, is transformed as a total sum into partial sums su m (n) over. Finally, the maximum of the channel levels is also determined group-related according to Eq. (20) using Pseudo Code 2, resulting in Dmax. m (n) is therefore an individual value for each audio compression channel, where all these maxima are the same within a common group. The introduction of the sum vector s mThis implies that all sum components must be detected individually and then transformed into the logarithmic domain, which increases the computational effort accordingly. With the described adjustments regarding channel-specific sums and maxima, and applying these to all corresponding equations, Eq. (16) for the hard-knee case becomes: − ∙ max{ − ∙ F( − ∙ ... for ^^(^) − ^^(^) > ^ / 2 (18.2b) or for ^^(^) − ^^(^) < −^ / 2 The automatic threshold in Eq. (21) is expressed by ^ ^^ (^) = ^^^{^^^{^ ^^^^ (^) + ^, ^^}, ^^} (21b) Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO As already indicated, this results in a significant increase in effort, but it offers the possibility of running many group processing operations in parallel through partitioning. With M unlinked channels, up to M parallel mono compressors can also be configured using manual thresholds. Figure 8 illustrates the described data organization using the example in Figure 5. For the linking vector vm = (0,1,1,0,1,1,1,0,0,1,1,0), the virtual channel sum s m (sum), the lower limit su m (“sum bound”), the sum level S m (sum level) and the maximum channel level Dmax mThe (max channel level) is listed. s1, s2, ..., u1, u2, ..., S1, S2, ..., M1, M2, ... are arbitrary numerical values. The crucial factor is which channels are assigned the same values. Operating and Display Concept: One possible, though by no means the only, way to operate this multi-channel compressor will be presented below. This assumes that the multifunctional multi-channel compressor is part of a mixing console system. Fig. 9 shows an operating section of a console system 1000 with a group processing unit in the form of a multifunctional multi-channel compressor. It is assumed that only adjacent channel strips or adjacent channels can be linked to groups for group processing. There is either a physical control unit in the form of parallel channel strips 180-1, … 180-M with their controls in the form of faders 190-1, …- 190-M or a virtual mixing console view on a screen.Each channel strip 180-1, … 180-M comprises two control buttons 205-1, …, 205-M; 206-1, …, 206-M: One control button 205-1, …, 205-M, referred to here as the activation button, activates or deactivates the corresponding audio compression channel in the multi-channel compressor for group processing. When deactivated, its influence I and weight W are set to zero. The corresponding audio signal passes through the audio compression channel without compression. The other button 206-1, …, 206-M, the link button, links each corresponding audio compression channel to its left neighbor (variant 1) or, in an alternative variant 2, to its right neighbor (variant 2). The choice of variant is irrelevant here. A set of link buttons essentially corresponds to the physical configuration of a link vector in Table 2.As soon as a link exists between two adjacent audio compression channels, group processing is already in place. Further audio compression channels can be added by chaining them using their link buttons 206. This can be indicated in a virtual view, for example, by framing the linked channels or changing the background area or its color. A separate view in the form of an embedded window or another page in the screen display must exist to define the operating mode and all relevant parameters. This can be done either by selecting the aforementioned frame or area, or by selecting a position of a selection button that forms a compression mode selection device. In Fig.Channel 9 shows three middle channel strips, 190-3 to 190-5, grouped together for processing by two left-facing Link buttons, 206-4 to 206-5. Signal processing is activated for all three channel strips, 190-3 to 190-5, in the audio compression channels 210-3 to 210-5. The activation buttons, 205-3 to 205-5, are active (black button position). The channel strips are shown schematically only with their output faders 190, 190-1, ... 190-M, and not with their full range of functions. This operation allows neighboring channels to be spontaneously grouped and subjected to common dynamic processing in one of the operating modes. The user should be able to quickly grasp the configuration if the channels are labeled accordingly. The arrangement of the control buttons 205, 206 can be arbitrary. Applicant: STAGETEC GmbH, September 12, 2025. Our reference: P19.090WO Alternative Implementation Based on Parallel Identical Processing Units in Analog or Digital Technology. In the last section, a simple two-step method for determining partial sums was introduced. This section presents another implementation variant that can also be realized in analog technology. The underlying principle is to build the multifunctional multi-channel compressor 200 as a chain of parallel processing units as basic elements 1100 (Fig. 10) to form one or more automixers and / or group compressors. This is described below starting from an analog signal path. Mathematically speaking, the functionality of the chained basic element 1100 corresponds exactly to the method described in the last section. Arrows pointing outwards at the connections indicate the presence of an active driver stage. Each basic element 1100-m comprises one input 220-m and one output 250-m.Furthermore, there is a link input 221-m for a binary link signal, which indicates whether the corresponding basic element is linked to an adjacent basic element or not. The incoming link information 1103-m is also communicated to the left-adjacent basic element 1100-m-1, as indicated by the dashed arrows. In the representation in Fig. 10, it should be noted that the basic element 1100-1 has no left-adjacent element. A basic element 1100-M (not shown) accordingly has no right-adjacent basic element. Each basic element 1100-m outputs a forward sum 1140-m via a forward sum output 1142-m to a forward sum input 1141-m+1 of the next linked right-adjacent basic element 1100-m+1. Likewise, it receives a backward sum 1150-m+1 of the right-adjacent basic element 1100-m+1 at a backward sum input 1151-m, which the latter outputs at a backward sum output 1152-m+1.Furthermore, all basic elements 1100 are linked to a common potential bar 1200, which is raised by the basic element 1100 detecting the highest level to a maximum potential corresponding to the highest level detected at the basic elements 1100 of the group. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO Each basic element 1100-m thus has, in addition to the input 220-m and the output 250-m, a potential bar input 1201-m and a potential bar output 1202-m, a forward summation input 1141-m and forward summation output 1142-m, as well as a reverse summation input 1151-m and a reverse summation output 1152-m. Various configurations are possible, two of which are explained in more detail here. Variant 1: A first variant is shown in Fig. 11. Shown is such a basic circuit element 1100 with index m.This element contains link information 1103-m indicating whether this element and the following one are linked to their left neighbor. If so, the corresponding link switches 1145-m are closed. The forward sum m-11140-m-1 includes all filtered and weighted, i.e., pre-processed, signal components of the left side. If the current element m 1100-m is linked to the left (link bit m is logically "1"), its optionally weighted and / or optionally filtered channel signal m 1110-m is added to the received forward sum 1140-m-1. This occurs in the summing unit 1130-m and results in the forward sum 1140-m that is passed on. A filter and a weighting device are indicated in block 1120. The forward sum m 1140-m results, which is then passed on to the right neighbor 1100-m+1.If the current element m 1100-m is not linked, the forward sum m 1140-mund and the current, pre-processed channel signal m 1110-m are identical, i.e., the forward sum m-11140-m-1 is not added. This is selected by a switch 1145-m, which is switched depending on the linking information 1103-m. If the link information 1103-m is zero / false, there is no left-hand link and switch 1145-m is open. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO If another basic element m+11100-m+1 is connected and linked (link bit m+1 is logical "1"), a reverse sum m 1150-m corresponds to the reverse sum m+1 1150-m+1, since a corresponding further switch 1155-m selects this sum (link bit m+1 is logical "1"), which is switched depending on the link signal forwarded by the right basic element 1100-m+1 or the link information 1103-m+1.The further switch 1155-m is in position B if the linking information of the right-adjacent basic element 1100-m+1 is true (linked to the basic element m), otherwise it is in position A (basic element 1100-m is the last basic element 1100 to the right of the grouping). The reverse sum m+1 1150-m+1 includes not only the sum of the right-hand side, but of all linked audio compression channels, i.e., the left, right, and current audio compression channels, because the last chained basic element 1100 on the right-hand side has the entire sum, and this sum is passed back recursively as the reverse sum 1150. The reverse sum m+11150-m+1 is therefore equal to the previously introduced virtual total sum s. m1350-m and is therefore routed to a detector 1160-m, which calculates a logarithmic level 1170-m. If no further circuit element m+11100-m+1 is connected, the end is reached, and the sum of the forward sum m-11140-m-1, if present and linked, and the current pre-processed channel signal 1110-m is equal to the backward sum m 1150-m and, as already mentioned, also simultaneously the total sum s mThe summing amplifiers 1130-m shown can be implemented using known operational amplifier circuits. Furthermore, a potential rail 1200 exists for all linked basic elements 1100, which the audio compression channel 210 with the highest level can pull to its maximum value. Known driver circuits with diodes can be used for this purpose; in the illustrated embodiment, this is part of a compression channel detector 1195-m, which determines the channel level 1210-m. The adaptive threshold 1220 is derived from this maximum of all linked channel levels 1210. Another switch 1205-m, which depends on the linking information 1103-m of the basic element 1100-m, ensures the correct linking of the basic elements 1100 in the grouping with the potential rail 1200. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO The channel gain calculation device 1250-m calculates the target gain for the audio compression channel 210.The calculation incorporates the logarithmic level 1170-m of the summed signal and, depending on the compression mode, a threshold value 1260-m as well as an optionally selectable / adjustable influence factor 1270-m. A selection circuit 1240-m chooses the appropriate value for the threshold value 1260-m depending on the operating mode. In "normal" group compression, this is a preselected / adjustable constant threshold value 1280-m; in group compression with an adaptive threshold, it is the maximum channel level 1218 added with an offset 1290-m; and in automix mode, it is the detected channel level 1210-m. The output amplifier 240-m is controlled by the target gain signal 1258-m. The lower limit sum 1340-m for the automixer in RMS mode is only indicated in the circuit diagram for space reasons and can be summed according to the same principle as the virtual sum sm 1150-m. The measured power is summed and serves as the power sum limit.The summation can be calculated analogously to the forward and backward sums, so that only linked channels are summed. Variant 2: As described, Variant 1 in Fig. 11 requires the linking information 1103-m+1 of the following element m+11100-m+1. Variant 2 in Fig. 12 can dispense with this information and the corresponding bit line, but requires three summing elements 1130-m, 1180-m, 1190-m instead of one. In this variant, the backward sum m+1 1150-m+1 corresponds only to the sum of the channels on the right-hand side and not, as before, to the grand total. If no further element 1100 is connected, this backward sum 1150-m+1 is equal to 0 (no signal), for example, by pulling the potential to signal ground via an electrical resistor.The reverse sum m 1150-m is obtained, as shown, simply by summing the current pre-processed channel signal m 1110-m and the reverse sum m+11150-m+1 at the summing unit 1180-m, if a left link exists (link bit m, i.e., the link information 1103-m is logically "1"). Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO The forward sum m 1140-m is obtained in the same way as in variant 1, while the final sum s. mThe summation is formed by summing the forward sum m 1140-m and the backward sum m+11150-m+1 at the summing mixer 1190-m, since this forward sum m 1140-m includes all relevant left audio compression channels 210-i (i ≤ m) including the current audio compression channel m, and the backward sum m+11150-m includes all relevant right audio compression channels 210-k (k > m), starting with m+1. In this variant 2, the connections via the switches 1145-m, 1155-m, 1205-m with the left basic element m-1 depend on the connection information 1103-m. In both variants, the illustrated gain block includes the calculation of the target gain and the smoothed gain, which is finally applied to the output amplifier 240-m, designed as a VCA. This part has already been fully described from a mathematical perspective in the previous sections.Stereo variant of the multi-channel compression method with envelope detector (peak detector). So far, only mono sums have been considered. Conventional, envelope-based bus compressors with stereo inputs and outputs often operate on the principle that the maximum of the detector signals from the left and right input channels is used as a single, combined detector signal, and both channels always undergo the same compression. This results in no shifts in the stereo image, and the currently louder channel, i.e., left or right, determines the compression behavior. If, as before, a virtual mono sum of the multi-channel compressor controls a number of input channels, and these are then panned and summed to a stereo signal, the result, with the same compressor settings, is generally different than if the stereo sum had been subjected to the described bus compression only after summing.The mapping function of the panorama control also plays a role here. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO The differing control behavior may not pose a problem in many cases, especially since no shifts in the stereo image occur in the case of the mono sum. However, if one wants to achieve exactly the same control behavior as in the case of the bus compressor, the following variant is recommended. Fig. 13 again shows a simple mixing console system into which the multi-channel compressor 200 is integrated. Mono channel strips 180-m are panned in a panorama unit 1600 and mixed directly to a stereo master bus 1700 at summing mixers 1710 and 1720. As shown, the sends 1510-m to effects units 550 are routed after compression, i.e., also after an output fader 190-m.The complete positions of the panorama controls 1610-m in this arrangement are transmitted as information to the multi-channel compressor, for example, as panning information 1620. Control channel panorama controls 346-m also exist in the detector path, which are set using the panning information 1620 in the same way as the panorama controls 1610-m in the panorama unit 1600 in the master bus 1700. The mono control channel signals 360-m are thus converted into stereo signals 361-m, 362-m, which are mixed to form a virtual stereo sum 371, 372. The detector device 380 therefore has two detectors for determining a right and a left sum level and, if necessary, two detectors for each control channel to determine a right and left channel level. When all weights w. iWith a gain of 1 and the filters also set to neutral, the virtual sums in the detector path of the compression control unit are equal to the respective summed portions of the channels in question, which are combined in the master bus. In this case, a situation has been created where the multi-channel compressor accesses the same signal as a hypothetical bus compressor accesses the sum of the input signals, assuming only these channels were mixed onto the bus. If an identical sidechain filter is used for all channels, this correspondence would still exist if the hypothetical bus compressor used the same sidechain filter, but only for its two input channels (L / R) instead of the entire set of input channels. Only when the weights are changed do different relationships result, which may be desirable. (With the help of the copied applicant: STAGETEC GmbH 12.09.)2025Our reference: P19.090WOPannings in the detector path make it possible to emulate a bus compressor using the multi-channel compressor. For the sake of simplicity, the following description is formulated only for global group processing. Partitioning to multiple parallel group processing operations proceeds in the same way as before. Panning a mono input signal is simply a dual mapping of the signal to two output channels in the form of an angle-dependent weighting. The weighting functions for an angle θ are given for the left and right channels by PL(θ) and PR(θ). m Let the panoramic angle for the m-th input channel be . Then, from the filtered and weighted input signals according to Eq. (2), the following stereo tuples (cL) result (cL). m (n),cR m (n)) through panning: This time the virtual signal sum is a stereo tuple (s L (n),s R (n)) and replaces Eq. (3): ^^(^) = ∑^ ^ ^^ ^^ ^ (^) (3.1b) ^^(^) = ∑ ^ ^ ^^ ^^ ^ (^) (3.2b) The individual components sL(n), sR(n) of the stereo sum and the individual components cL m (n), cR m (n) of the acquired stereo channels now pass separately through corresponding envelope detectors. Subsequently, the maximum of the left and right components is determined for each channel and the sum: ^^(^) = max {^ ^ ^^^{^^ (^), {^^^{^^ (^)}} (8b) ^^(^) = max {^^{^^(^)}, ^^{^^(^)}} (9b) The combined detector signals thus obtained d m (n) and s e(n) for the stereo case serve as the basis for the further process steps of the method as before and, in this role, replace the original detector signals in Eq. (8) and Eq. (9). The stereo applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO The method is limited to group compression with envelope generators with manual or automatic threshold. The methodology is less suitable for the application of the automixer, and RMS detection is also omitted. Up to now, stereo channels have been obtained by panning. In the case of true stereo input channels, these can be fed directly to the detector by simply mixing the left input channel to the left sum and the right channel to the right sum. The described multi-channel compressor based on the principle of a virtual sidechain bus can be considered an independent category.This compressor operates on the principle that while channels are compressed together using an internal sum and thus influence each other, they remain as individual signals and can be processed further. This makes the multi-channel compressor an alternative to a conventional bus compressor. Whether integrated into a mixing console setup or inserted externally as a standalone device via post-fader insert paths into all relevant channels, it solves the classic problem of varying wet / dry mixes when using effects units, without requiring these units to be on a shared bus. In this configuration, the sends are placed post-fader after the compressor. Furthermore, the ability to weight channels and set individual ratios allows for processing techniques that are not possible with a conventional bus compressor.The presented compression with automatic threshold is also a novelty and can only be achieved because, as in this case, the compressor has access to all individual channels. The threshold is derived from the level of the loudest channel plus a constant offset. This mode can be seen as complementary to an automixer in a certain sense, because a channel with a high level pulls the threshold upwards, thus resulting in less compression for the other channels, while an automixer does the opposite by significantly attenuating low-level channels. This method can be applied to groups of singers or instruments. Thanks to the automatic threshold, the user no longer needs to set or manually adjust it. They only need to adjust the offset and ratio appropriately.The multi-channel compressor operates on a mono sum, even if the relevant channels are later distributed in the stereo panorama and mixed to a bus. However, to achieve the same control behavior as a peak-based bus compressor, which uses the maximum signal levels of the left and right channels in the detector path, another variant was introduced. In this variant, based on provided panning information, the multi-channel compressor copies the resulting stereo sum of a master channel and can thus exhibit essentially the same control behavior as a bus compressor on the master channel itself. This assumes that the multi-channel compressor is integrated as a block into a simple mixing console system where channels are always summed to a stereo master bus—a fairly common configuration.A sequential implementation has been described in which linked neighboring channels can be spontaneously combined into automixes and group compressions and displayed as such, without requiring the creation of separate buses. Separate views on an integrated or connected screen or display are required for compression settings. The restriction to neighboring channels can be mitigated if the channel strips have source selection, i.e., if an input router exists in the system. With the described two-step method for sum and maximum, the overall computational effort remains approximately constant, regardless of how the groups are partitioned.The increased effort required for all calculations related to the channel sum is justified by the fact that, with this implementation, the multi-channel compressor can perform up to M / 2 non-overlapping group compressions or up to M individual channel compressions when needed, assuming an even number of M input channels. The same procedure, as described, can also be implemented using parallel basic blocks that have the same controls for grouping adjacent channels and can even be implemented with analog signal processing. This is made possible by a forward and reverse sum that is passed between adjacent modules. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO Overall, the multifunctional multi-channel compressor with at least one group compressor with adaptive threshold for signal processing is available, which can optionally be further enhanced with a virtual bus compressor or...a dynamic processor can be implemented in the audio channel control unit, which can perform both automixing and compression of individual signals and groups alike, and provides a powerful, flexible tool for audio processing.

[0002] Applicant: STAGETEC GmbH 12.09.2025 Our Ref: P19.090WO Reference 1 Signal processing unit 11, 12, 13 Channels 20 Summer 30 Bus signal 41, 42, 43 Send channel 51, 52, 53 Weighting unit 60 Summer 70 Sum signal 80 Effects unit 90 Effect signal 100 Bus compressor 110 Bus signal 120 Additional summmer 130 Output signal 180, 180-1, 180-2, 180-M Channel strips 190, 190-1, 190-2, … 190-M Output fader 200 Multi-channel compressor 205, 205-1, …, 205-M Control knob 206, 206-1, …, 206-M Other knob / Link buttons 210, 210-1, 210-2, …, 210-M Audio compression channels 211, 212, 215 Individual tracks 220, 220-1, 220-2, … 220-M Input 221 Link input 240, 240-1, 240-2, …, 240-M Output amplifier 250, 250-1, 250-2, … 250-M Output 260, 260-1, 260-1, …, 260-M Output signals 270 Bus signal 280 Bus summer 290 Master bus summer 300 Compression control device 310, 310-1, 310-2, …, 310-M Audio control channel 320, 320-1, 320-2, …, 320-M Input 330, 330-1, 330-2, …, 330-M Side channel filter Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO 340, 340-1, 340-2, …, 340-M Weighting device 341 Detector elements 342 Line component summing 343 Logarithmic element 345 Maximum finder 346-m Control channel panorama control 350 Channel gain calculation device 355 Summer 360 Audio control channel signals 361-m, 362-m Stereo signals 370 Sum control signal 371, 372 Stereo sums 380 Detector device 390 Target gain signal 510 Send mixes , 510-1, 510-2, …, 510-M 520 Weighting devices , 520-1, 520-2, …, 520-M 550 Effects unit 560 Output signal 710-1 Group compression 810-1, … 810-M incoming signals 820 Router 830 Mixing console 840 Multi-channel compressor (200!) 910 Input data 920 Output data 930 Processor 940 Memory 1100 Basic elements 1103-m Linking information 1140-m-1 Sum 1110-m Channel signal 1120 Block 1130 Summer 1140-m Forward sum m 1141-m Forward sum input Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO1142-m Forward summing output 1145-m Switch 1150 Reverse summing 1151-m Reverse summing input 1152-m Reverse summing output 1155-m Additional switches 1160-m Detector 1170-m Logarithmic (summing) level 1180-m Summer 1190-m Summer 1195-m Compression channel detector 1200 Potential rail 1201-m Potential rail input 1202-m Potential rail output 1205-m Another switch 1210-m Channel level 1218 Maximum channel level 1220 Adaptive threshold 1240-m Selection circuit 1250 Channel gain calculation device 1258-m Target gain signal 1260-m Threshold 1270-m Influence factor 1280-m Constant threshold 1290-m Offset 1340-m Lower barrier sum 1350-m Total sum sm 1510-m Sends 1600 Panoramic unit 1610-m Panoramic controller 1620 Panning information 1700 Master bus 1701, 1702 Master bus stereo signals.

Claims

1. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO Patent claims 1.A multi-channel compressor with automatic threshold setting comprising a plurality M of audio compression channels (AKK-m), wherein each audio compression channel (AKK-m) includes an input (Em) and an output (Am) and a controllable output amplifier (compressor VCA-m) connected therein via an audio signal, and a compression control device (KSE) comprising, for each of the plurality M of audio compression channels (AKK-m), an audio control channel (ASK-m) whose audio control channel input is connected to the input of the audio compression channel (AKK-m), and an audio control channel output (StA-m) which is connected to a control input (StE) of the controllable output amplifier (VCA-m), wherein several of the plurality M of audio compression channels (AKK-m) are linked to form a group or groups, wherein the compression control device (KSE) additionally includes at least one summing device (SE) for forming audio control channel summation signals. (see p(n)) (AVS-p) of interconnected audio control channel signals (c m (n)) the majority of the audio control channels (AKK-m) for each of the groups and at least one detector device for determining channel levels (D m (n)) for the individual audio control channel signals and sum levels (Sp(n)) for the audio control channel sum signals (sp(n)), wherein the compression control device (CCD) has a channel gain calculation device (CCD-m) which calculates a target gain signal value for each audio compression channel as a function of the sum level (S p (n)), into which the audio control channel signal has been received, and an adaptive reference threshold is calculated, wherein attenuation of the gain occurs when the respective sum level exceeds the respective adaptive reference threshold, wherein the compression control device (CCD) has a maximum value determination device for determining the current maximum channel level (max m {D m(n)}) and comprises a comparison threshold calculation unit which, when generating the adaptive comparison threshold, adds a fixed or adjustable non-negative offset to the current maximum channel level. Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO2. Multi-channel compressor with automatic threshold setting according to claim 1, characterized in that the comparison threshold calculation unit comprises a limiting unit configured to limit the comparison threshold to the lower limit threshold when determining it, and / or to limit it to the upper limit threshold when determining it.

3. Multi-channel compressor with automatic threshold setting according to claim 1 or 2, characterized in that,that each of the audio control channels comprises a side-chain filter (SFC).

4. Multi-channel compressor with automatic threshold setting according to one of the preceding claims, characterized in that each of the audio control channels comprises a weighting device (Wm).

5. Multi-channel compressor with automatic threshold setting according to one of the preceding claims, characterized in that the signals representing the target amplifier values ​​determined in the channel gain calculation device (KVBE-m) are passed through a smoothing device before being used to control the controllable output amplifiers.

6. Multi-channel compressor with automatic threshold setting according to one of the preceding claims, characterized in that the determination of the values ​​is carried out in a multi-stage iterative or multi-stage parallel process.wherein in an iteration one or more of the magnitude and / or one or more intermediate results for one of the audio compression channels (AKK-m) are determined, wherein in an iteration different magnitudes or different intermediate results for different audio compression channels (AKK-M) can be determined simultaneously, wherein the iterations of a stage are executed a number corresponding to a number of the plurality M of audio compression channels (AKK-m). Applicant: STAGETEC GmbH 12.09.2025 Our reference: P19.090WO7. Multi-channel compressor with automatic threshold setting according to one of the preceding claims, characterized in that the audio compression channels (AKK-m) are indexed and, via the linking device, the respective audio compression channel (AKK-m) can only be linked to an audio compression channel (AKK-m-1) adjacent with respect to indexing or not linked.wherein all audio compression channels (AKK-m) can be linked either to an audio compression channel (AKK-m) with a smaller index or all only to an audio compression channel (AKK-m) with a larger index.

8. Multi-channel compressor with automatic threshold setting according to one of the preceding claims, characterized in that each of the audio compression channels (AKK-m) is assigned a linking device for linking to at least one other audio compression channel (AKK-m), such that the links of the plurality M of audio compression channels can be determined, wherein each audio compression channel can only be linked to one of the groups.

9. Multi-channel compression method with automatic threshold setting for a plurality M of audio signals, comprising the steps: feeding audio signals into a plurality M of audio compression channels,wherein each of the audio compression channels has an input and an output as well as a controllable output amplifier arranged between them; determining the individual gains for the output amplifiers of the individual audio compression channels of a group or several groups by forming a summed signal from the audio signals of audio compression channels grouped together in a side-channel processing operation and determining a summed level based on the summed signal, and making the control signals representing the gains for the individual audio compression channels by comparing the summed level into which their audio signals have been incorporated with a threshold value, and wherein an attenuation of the gain occurs when the summed level exceeds the threshold value, wherein the threshold value is determined adaptively for each channel group.by determining a channel level for each audio compression channel of a group and, for the purpose of calculating the adaptive threshold, determining a maximum channel level of the group and increasing this by a positive offset, wherein the adaptive threshold is used for all audio compression channels of the respective channel group.

10. Method according to claim 8, characterized in that the following additional method step is performed: defining links between the individual audio compression channels to form one group of channels or several groups of channels.

11. Method according to claim 8 or 9, characterized in thatthat an upper limit threshold and / or a lower limit threshold for the adaptive threshold is set or detected, and that the automatic threshold determined based on the channel levels and the positive offset is limited downwards by the lower limit threshold and / or upwards by the upper limit threshold.

Images