Method and device for adjusting the loudness of an audio signal and motor vehicle audio system comprising the device

The dual-rate loudness adjustment method in audio systems addresses undesirable volume fluctuations by quickly responding to changes in audio sources or tracks, maintaining smooth playback and reducing the need for user adjustments.

DE102012014310B4Active Publication Date: 2024-11-21E SOLUTIONS
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
DE102012014310
Authority / Receiving Office
DE · DE
Patent Type
Patents
Current Assignee / Owner
Filing Date
2012-07-19
Publication Date
2024-11-21
Estimated Expiration
2032-07-19

AI Technical Summary

Technical Problem

Existing audio systems, particularly in motor vehicles, experience undesirable volume fluctuations due to changes in audio signal sources or tracks, which are not attributed to the dynamics of the audio information, leading to uncomfortable listening experiences.

Method used

A method and device for adjusting loudness in real-time by using a dual-rate loudness adjustment mechanism, where the loudness is increased or decreased at different rates to maintain original dynamics while quickly addressing large jumps, involving automatic gain control and silence detection.

Benefits of technology

The solution effectively minimizes abrupt volume changes, ensuring a smooth and comfortable listening experience by rapidly adjusting to loudness variations without user intervention, while preserving the audio's original dynamics.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000000_0000_ABST
    Figure 00000000_0000_ABST
Patent Text Reader

Abstract

Method (500) for adjusting the loudness of an audio signal, comprising the steps: Analyzing (504, 506, 508) a loudness of the audio signal with respect to a target loudness and determining a gain factor for raising or lowering the loudness of the audio signal to continuously raise the loudness of the audio signal to the target loudness at a first rate or lower it to the target loudness at a second rate; Analyzing (510) a resulting loudness of the audio signal scaled on the basis of the determined gain factor with respect to exceeding a loudness limit; and If the resulting loudness exceeds the loudness limit, reduce (512) the loudness of the audio signal at a second rate higher than both the first rate and the second rate.
Need to check novelty before this filing date? Find Prior Art

Description

Technical area

[0001] The present disclosure relates generally to the field of audio signal processing. Specifically, a method and a device for adjusting the loudness of an audio signal are disclosed. A motor vehicle audio system comprising the device is also disclosed. background

[0002] Volume fluctuations during audio playback are generally undesirable when they are not due to the dynamics of the audio information being reproduced (such as a music track) but rather to purely technical effects. For example, switching the signal source with significantly different levels of the output audio signal will result in an unwanted volume jump during playback. A good example of this is switching from a radio receiver to a compact disc (CD) player as the signal source. Similarly, within a single audio signal, an unwanted volume jump can occur when transitioning from one track to the next if the tracks were recorded or saved with different settings.

[0003] To quantify differences in perceived loudness, a measure called "loudness" was introduced. Loudness is defined in various standards and generally has a proportional relationship to human loudness perception. Therefore, loudness doubles when a reproduced audio signal is perceived as twice as loud.

[0004] Besides methods that determine the actual loudness of a real sound event, there are also those that merely assess the relative loudness of different audio signals to one another. Such methods can be used, in particular, to equalize the loudness of audio signals without knowledge of other factors influencing the listener's perception of loudness—for example, the efficiency of the amplifiers and loudspeakers used, as well as the listener's distance from the loudspeakers. An example of this is the so-called ITU loudness, described in the standard BS.1770-2 (03 / 2011) of the International Telecommunication Union, Radiocommunication Sector (ITU-R). In addition to this ITU-R standard, other standards and models exist for calculating loudness.

[0005] Given an audio signal, its loudness can be influenced by scaling during playback. Scaling with a value or factor less than 1 generally reduces the signal's loudness, while scaling with a value or factor greater than 1 generally increases it. The relationship between scaling and loudness change depends on the audio signal itself and the standard or model used to calculate loudness.

[0006] If the loudness of an audio signal needs to be approximated to a desired value, this can be achieved through appropriate scaling, provided the relationship between scaling and loudness change is known. If the signal is fully available before playback (e.g., in the form of a music file), a constant gain factor can be calculated, after which the average loudness of the audio signal will correspond to the specified target value.

[0007] In many applications, such as automotive audio systems, offline analysis and loudness adjustment are not possible, for example, because the entire signal is not available before playback (e.g., with radio reception). In such cases, the analysis and adjustment must occur in real time. Several constraints must be considered here. Generally, in many real-time applications, the delay of the audio signal during its analysis and adjustment should be kept as low as possible. Furthermore, the original dynamic range of the audio information should be preserved as much as possible. For this, a relatively slow loudness adjustment is desirable. On the other hand, it is necessary to absorb large loudness jumps—for example, when switching the signal source. The latter requires a fast loudness adjustment.

[0008] To prevent unwanted loudness jumps, US 2008 / 0253586A1 teaches the monitoring of an audio signal for transitions between two signal segments of differing loudness (e.g., between program playback and a subsequent commercial break). Furthermore, a long-term loudness is determined for a first segment of the audio signal, and a short-term loudness is determined for a second, shorter segment. Once a segment transition is detected, the gain factor for the audio signal is adjusted in two stages. The gain factor is determined immediately after the segment transition for a predetermined period based on the short-term loudness. After this period, the gain factor is determined based on the long-term loudness.

[0009] In a similar context, US 2010 / 0272290A1 proposes determining the gain factor by considering the loudness before and after the segment transition. Long-term and short-term loudness are again taken into account when determining the gain factor.

[0010] DE 10 2010 003 306 A1 relates to a volume control device for regulating the volume of different audio signals, for example, different audio signals from different radio stations. The device comprises a volume control for automatically adjusting the volume of the selected audio signal and a storage medium for storing volume control information for the individual audio signals. Such volume control information includes at least one state variable of the automatic volume control or directly a volume control variable. The volume control information is stored in the storage medium in such a way that when the previously selected audio signal is deselected, the current state variable or current volume control variable is stored in the storage medium as volume control information for the previously selected audio signal.When an audio signal is selected, the volume control information stored in the memory for the respective audio signal is used to control or adjust the volume of the selected audio signal. Brief overview

[0011] The present disclosure aims to provide an efficient method and device for loudness adjustment. In particular, at least in some implementations, it should eliminate the need for users of an audio system to repeatedly and undesirably readjust a volume control to compensate for sudden loudness changes (such as when switching sources or tracks). This requires, on the one hand, largely preserving the original dynamic range of the audio signal, and on the other hand, reliably preventing undesirably large loudness jumps.

[0012] According to the invention, a method, a device, and a motor vehicle audio system are provided as defined in the independent claims. Developments are described in the dependent claims.

[0013] The loudness of the audio signal is reduced at the second rate if the resulting loudness exceeds the loudness limit. If, however, the resulting loudness does not exceed the loudness limit, the audio signal can be continuously adjusted to the target loudness by raising or lowering it at the first or second rate (possibly with repeated adjustments to the gain factor). The loudness of the audio signal can be selectively increased or decreased.

[0014] When reducing the loudness of the audio signal at the second rate, the audio signal is preferably brought down to the loudness limit. However, it would also be conceivable to reduce the loudness of the audio signal at a loudness value other than the loudness limit when reducing it at the second rate.

[0015] The reduction of the loudness of the audio signal at the second rate can occur in a single step (i.e., instantaneously) or, preferably, continuously. Within the scope of this disclosure, a continuous loudness adjustment can, for example, be performed in several discrete steps. In other words, a continuous loudness adjustment can extend over a certain time period.

[0016] To reduce the loudness of the audio signal at the second rate, the determined gain factor is preferably modified. Subsequently, the audio signal is preferably scaled based on the modified gain factor. The degree of modification can depend on the resulting loudness of the audio signal scaled with the (unmodified) gain factor or be fixed.

[0017] The target loudness and loudness threshold can be variable parameters. The target loudness can be predefined by the system, by a user, or adjustable at runtime. Preferably, the loudness threshold is derived from the target loudness. The loudness threshold is preferably set above the target loudness by a fixed or variable tolerance value.

[0018] In one implementation, a further, user-adjustable scaling stage, corresponding to a conventional volume control, can be added. In this case, the target loudness can be a constant value at runtime; however, additional user adjustment of the target loudness and / or the loudness threshold is possible.

[0019] The rate used to increase or decrease loudness can generally refer to the speed at which the scaling changes to affect (increase or decrease) the loudness of the audio signal. Thus, the rate can generally characterize the rate at which the gain is adjusted to achieve the desired loudness (e.g., the target loudness and / or the loudness threshold). The change in scaling can be specified in dB / s or another unit.

[0020] Preferably, the first rate for increasing loudness or the second rate for decreasing loudness is in the range of approximately 0.1 dB / s to 2 dB / s. The second rate for decreasing loudness is preferably in the range of approximately 0.1 dB / ms to 2 dB / ms. Independently or additionally, the first rate for increasing loudness or the second rate for decreasing loudness preferably results in a control time of approximately 5 s to 5 min, and / or the second rate for decreasing loudness preferably results in a control time of approximately 1 ms to 1 s. The control time here refers to the period until a desired loudness is substantially reached.

[0021] The initial rate for increasing loudness preferably differs from the initial rate for decreasing loudness. The decrease preferably occurs faster than the increase.

[0022] Analyzing the loudness of the audio signal with respect to the target loudness and determining the gain factor based on this analysis is preferably performed in conjunction with automatic gain control (AGC). The automatic gain control may include a feedback loop. Alternatively, particularly if the relationship between the choice of gain factor and the resulting loudness change is known, the automatic gain control is preferably of the feedforward type.

[0023] Analyzing the loudness of the audio signal with respect to the target loudness can include a first loudness measurement. Analyzing the audio signal scaled by the determined gain factor can include a second loudness measurement that reacts more quickly to changes in the input signal than the first measurement, for example, by using shorter averaging times in the analysis. In this case, one can speak of a "short-term loudness" measurement. Preferably, the second analysis is designed as a peak detection method.

[0024] Preferably, the audio signal is delayed. This delay is performed to synchronize the audio signal to be played with a reduction in loudness, preferably at the second rate.

[0025] Preferably, silence detection is performed for the audio signal. If silence is detected, preferably at least one process step is omitted, for example, adjusting the gain factor to achieve the target loudness and / or scaling the audio signal and / or analyzing the resulting loudness of the audio signal scaled with the determined gain factor and / or reducing the loudness of the audio signal at the second rate.

[0026] Preferably, the audio signal is based on a radio signal. In the case of poor radio reception, preferably at least one of the following process steps is omitted: adjusting the gain factor to achieve the target loudness and / or scaling the audio signal, analyzing the resulting loudness of the audio signal scaled with the determined gain factor, and attenuating the loudness of the audio signal at the second rate. Poor radio reception can, for example, be indicated by a radio receiver.

[0027] In connection with the execution of the present method, the audio signal is preferably sampled. The audio samples can then be processed in blocks. The analysis of the loudness of the scaled audio signal is preferably based on a smaller sample block size than the analysis of the loudness of the unscaled audio signal. Alternatively or additionally, an adjustment of the gain factor at the second rate can be based on a smaller sample block size than the adjustment of the gain factor at the first rate or the second rate. Brief description of the drawings

[0028] Further aspects, advantages, and details of the present disclosure will become apparent from the following description of exemplary embodiments and from the figures. These show: Fig. 1A and Fig. 1B Block diagrams of exemplary implementations of a motor vehicle audio system; Fig. 2 a block diagram of an embodiment of a device for adjusting loudness in the audio system according to Fig. 1A or Fig. 1B; Fig. 3 components of a first functional block of the device according to Fig. 2; Fig. 4 components of a second functional block of the device according to Fig. 2; Fig. 5 a flowchart of an embodiment of a loudness adjustment method as described by the second functional block according to Fig. 4 is carried out; Fig. 6 to 8 diagrams illustrating the loudness setting according to the present disclosure for a test signal; Fig. 9 diagrams illustrating the loudness setting according to the present disclosure for a sequence of three musical pieces with increasing loudness; and Fig. 10 diagrams showing the loudness setting for the same sequence of music tracks as Fig. Show 9 without quick volume adjustment. Detailed description

[0029] The present disclosure is explained below with reference to schematic block diagrams. It is understood that the technical principles underlying the block diagrams can be implemented as hardware, software, or a combination of both. For example, possible implementations may employ digital signal processors (DSPs), application-specific integrated circuits (ASICs), and similar components.

[0030] The block diagram according to Fig. Figure 1A illustrates an embodiment of an audio system 10 for installation in a motor vehicle. The audio system 10 comprises a variety of signal sources, such as a CD player 12, a radio receiver 14, and an external input 16, for example, for portable music playback devices. Further signal sources implemented in this embodiment include a telephony module 18 (for example, with hands-free functionality) and a navigation module 20 for outputting route guidance instructions. It is understood that neither the telephony module 18 nor the navigation module 20 need necessarily include all telephony and navigation functionalities. Rather, the telephony module 18 can have a suitable interface for communicating with a motor vehicle telephone system (not shown). Similarly, the navigation module 20 can include an interface to a motor vehicle navigation system (not shown).

[0031] The in Fig. The audio system 10 shown in Figure 1A further comprises a device 22 for adjusting the loudness, hereinafter also referred to as the loudness control 22. The loudness control 22 is coupled on its input side to the signal sources 12, 14, 16, 18, 20 already described. If necessary, a signal selection device (not shown) can be provided between these signal sources 12, 14, 16, 18, 20 and the loudness control 22, which ensures that the loudness control 22 receives an audio signal from only one of the signal sources 12, 14, 16, 18, 20.

[0032] The audio system 10 further comprises a power amplifier 24, which is intended to amplify an output signal of the loudness control 22. The power amplifier 24 is coupled to a loudspeaker arrangement 26 in a known manner on its output side.

[0033] The loudness control 22 can also be followed by a further, user-adjustable scaling unit 28 in the sense of a conventional volume control. Likewise, the audio system 10 can be designed such that only a selection of sources (such as the CD player 12, the radio receiver 14, and the external input 16) are fed to the loudness control 22, while other sources (such as telephony 18 or navigation 20) are fed directly to the power amplifier 24 or further processing stages. Such an embodiment is shown in Fig. Figure 1B is shown, with an additional source-specific scaling provided.

[0034] It goes without saying that the representations are in accordance with the Fig. 1A and Fig. Figure 1B is merely schematic and is not intended to provide a complete description of a motor vehicle audio system. Therefore, further functional blocks, for example for source-specific equalization of the input signals and for adjusting the output signal of audio system 10, are not shown.

[0035] Fig. Figure 2 shows a block diagram of an embodiment of the loudness control 22 for processing a stereo audio signal. It should be noted that the arrangement can also be designed for other audio channel configurations, such as for processing a 5.1 or 7.1 signal. The loudness control 22 comprises a total of four functional blocks, labeled A, B, C, and D.

[0036] The first functional block, A, is configured to analyze the loudness of an input audio signal relative to a predefined target loudness and, based on this analysis, determine a gain factor for increasing or decreasing the loudness of the audio signal. The second functional block, B, is configured to analyze the resulting loudness of the audio signal scaled by the gain factor determined in functional block A. This analysis checks whether the resulting loudness exceeds a loudness threshold derived from the target loudness. The third functional block, C, is then responsible for the actual increasing or decreasing of the loudness. The loudness of the audio signal is continuously, i.e., not instantaneously, increased to the target loudness with a first rate or decreased to the target loudness with a second rate.However, if function block B determines that the resulting loudness exceeds the loudness limit, the third function block C reduces the loudness of the audio signal at a second rate that is higher than both the first rate and the second rate.

[0037] As in Fig. As shown in Figure 2, each of the three function blocks A, B, and C is equipped with a separate input for the two-channel audio signal. The fourth function block, D, which serves for silence detection, is only coupled to function block A at its input. The task of the fourth function block, D, is to examine the audio signal for silent sections. This can be achieved, for example, by comparing the measured loudness of the input signal with a threshold value. If such silent sections are detected in the audio signal, the control of the gain factor in function blocks A and B is prevented. For this purpose, function block D includes a silence detector 30 with a control output 32 each to function blocks A and B. Alternatively, function block D can also include an input for the audio signal itself and examine it directly for the presence of silence.

[0038] The following section explains in more detail the structure and function of functional block A with regard to loudness analysis and the determination of the amplification factor. As in Fig. As shown in Figure 2, functional block A includes a loudness detector 34 for measuring the loudness of the input signal. The loudness of the input audio signal can be determined using various standards or models. For example, the International Telecommunication Union, Radiocommunication Sector (ITU-R) standard BS.1770-2 (03 / 2011) can be mentioned in this context.

[0039] The loudness-representing output signal of the loudness detector 34 is fed to both the silence detector 30 of functional block D and an automatic gain control (AGC) 36 in functional block A. The silence detector 30 analyzes the output signal of the loudness detector 34—as described above—with respect to silent sections. The AGC 36, on the other hand, compares the current loudness of the audio signal in real time with the user-specified or predefined target loudness and determines a gain factor (hereinafter also referred to as "AGC Gain" or "Gain") based on this comparison.

[0040] Within the scope of this disclosure, the terms "gain" and "amplification factor" are used somewhat synonymously and generally refer to a parameter on the basis of which the audio signal is scaled. A gain of 0 dB corresponds to a scaling of 1.

[0041] It is important to note that AGC 36, depending on the result of comparing the loudness of the audio signal with the target loudness, either increases or decreases the loudness of the audio signal. If the loudness of the audio signal is below the target loudness, the gain factor is chosen such that the loudness of the audio signal is increased (i.e., the resulting scale from the determined gain factor has a value >1). Conversely, if the loudness of the audio signal is above the target loudness, the gain factor is chosen such that the loudness of the audio signal is decreased (i.e., the resulting scale from the determined gain factor has a value <1). The gain factor determined by AGC 36 in this way is output by function block A to both function block B and function block C.

[0042] In function block B, the gain factor obtained from function block A is first converted into a scaling value. For this purpose, function block B includes a converter 40. As in Fig. As shown in Figure 2, functional block B further comprises a multiplier 42 and a pre-filter 44 for each of the two channels of the input audio signal. The input signal is scaled by the multiplier 42 using the scaling value obtained from the converter 40. The scaled input signal is then processed by the pre-filter 44 for the purpose of a second loudness measurement in a detector 46 provided for this purpose.

[0043] Detector 46 is responsible for analyzing the loudness of the scaled audio signal in relation to the loudness limit. More precisely, detector 46 analyzes the loudness to determine whether the loudness limit is exceeded. The loudness of the scaled audio signal determined by detector 46, relative to the loudness limit, is output to a processing unit 48. If the loudness limit is not exceeded, no further steps are initiated by processing unit 48 in functional block B. The audio signal can then be amplified to the target loudness in functional block C based on the gain factor determined by AGC 36 and the corresponding AGC-determined rate.

[0044] On the other hand, if the detector 46 signals to the calculation unit 48 that the loudness limit has been exceeded, the amplification factor determined (initially autonomously) by the AGC 36 is immediately reduced. As in Fig. As illustrated in Figure 2, the processing unit 48 is coupled to the AGC 36 for this purpose. The reduction of the gain factor by function block B upon detection of an exceedance of the loudness threshold directly results in a lower gain of the audio signal than in the case of "standalone" operation of the AGC 36 (i.e., when the loudness threshold is not exceeded). Accordingly, the loudness of the audio signal is reduced at a rate higher than the rate originally provided for in function block A.

[0045] Function blocks A and B are configured to have different rates (or speeds) for adjusting loudness. For example, the rate for raising or lowering the scaling, and consequently the loudness of the audio signal, by the AGC 36 in function block A can range from approximately 0.1 dB / s to 2 dB / s (e.g., at 1 dB / s), while the rate for lowering the scaling by function block B ranges from approximately 0.1 dB / ms to 2 dB / ms (e.g., at 0.5 dB / ms). Alternatively or additionally, the rate of raising or lowering can depend on the deviation of the measured loudness from the target loudness or the loudness threshold itself, for instance, to ensure regulation within a defined time period regardless of the magnitude of the required scaling change.Accordingly, the rate for increasing or decreasing loudness in function block A can result in a control time of approximately 5 seconds to 5 minutes until the target loudness is reached. Conversely, the rate for decreasing loudness in function block B can result in a control time of approximately 1 millisecond to 1 second until the loudness threshold (or another desired value) is reached. Different rates for increasing and decreasing loudness can be used in function block A. In particular, the increase can be slower than the decrease.

[0046] The gain factor determined in function block A, optionally influenced by function block B, is provided to function block C. For example, function block C can access a memory area assigned to the AGC 36 (in Fig. (2 not shown) to read the current gain factor. Similarly, function block B can have read and write access to this memory area to influence the gain factor stored there by the AGC 36 when the loudness limit is exceeded, as described above.

[0047] The current gain factor read from function block C is first converted into a scaling value by a local converter 50 – as described above for converter 40 of function block B. As in Fig. As shown in Figure 2, functional block C further comprises a multiplier 52 for each channel of the input audio signal to scale the output signal based on the scaling value determined by the converter 50. Upstream of the multiplier 52, functional block C provides a delay 54 for each of the two channels. The delays 54 delay the audio signal input to functional block C by a fixed value, which is determined by the duration of the gain reduction by functional block B (i.e., in accordance with the reduction of the loudness of the output signal upon detection of an exceedance of the loudness threshold). The delays 54 remain active even if the threshold is not exceeded.

[0048] In a psychoacoustically advantageous manner, rapid adjustment is thus made in the direct path of the loudness control 18 (via function block C) to counteract any detected loudness jump in the output audio signal of function block C. If, however, the loudness remains below the loudness threshold, only a slow adjustment towards the target loudness occurs, corresponding to the inherent adjustment time of function block A. Such a slow adjustment is less noticeable and therefore not psychoacoustically disruptive. Combined, these psychoacoustic effects mean that the user is less frequently prompted to adjust the loudness by operating the volume control. At the same time, the operation of the loudness control 22 does not depend on knowledge or determination of track boundaries, source changes, or similar events.

[0049] To explain the control behavior of the AGC 36 and the functional block B in more detail, we will now refer to the Fig. 3 and Fig. 4 referenced. Fig. Figure 3 shows the essential components of the AGC 36 in a block diagram, while Fig. 4 also illustrates the essential components of function block B in a block diagram.

[0050] As in Fig. As shown in Figure 3, the AGC 36 includes an input-side adder 60. The adder 60 receives the current loudness, determined by the loudness detector 34, as its first input signal. The second input signal is the target loudness, which is specified or set by the system or user. The adder 60 adds the negative value of the current loudness and the target loudness. In other words, the current loudness is subtracted from the target loudness, and the corresponding loudness difference is calculated.

[0051] The calculated loudness difference is then fed into a sequence consisting of an adder 62, a multiplier 64, and another adder 66. This sequence of components 62, 64, 66 also receives the current gain factor y and a time constant. The current gain factor y is read from a memory area 68 assigned to the AGC 36. The value of the time constant depends, among other things, on whether the loudness difference determined by the adder 60 is positive (current loudness lower than the target loudness) or negative (current loudness higher than the target loudness). A positive loudness difference requires an increase in loudness, while a negative loudness difference requires a decrease in loudness. The increase and decrease in loudness are performed with different time constants (release time factor and attack time factor, respectively) and therefore result in different control times.

[0052] The calculation of the time constant for adjusting the scaling in the case of attenuation (i.e., reducing the loudness) for the sampled audio signal is performed using the following formula: atc=1−e−2.2⋅Bfs⋅AT where: atc = the attack time factor (time constant) to be applied B = Sample block size fs = sampling rate AT = Attack Time (control duration) in seconds

[0053] The calculation of the release time factor is analogous.

[0054] Based on the in Fig. The 3 illustrated input operations calculate the sequence of components 62, 64, 66, the gain factor x to be newly written to memory area 68 according to the following program instruction: lastGain=atc*targetGain+(1−atc)*lastGain where: lastGain = the gain factor x to be newly written to memory area 68 or the gain factor y read from memory area 68. atc = Attack Time factor. targetGain = target gain derived from loudness difference.

[0055] As in Fig. As shown in Figure 3, the gain factor g' read from memory area 68 can differ from the gain factor g last written to memory area 68 by the sequence of components 62, 64, 66. This difference is due to the possibility of external influence on the gain factor x written by AGC 36 by function block B, based on an influence value d output by a component of function block B to reduce the gain factor. As explained above, a reduction of the gain factor always occurs when a loudness threshold is detected in function block B.

[0056] In this context, it should be noted that the operation of the loudness control 22 is based on sampling the audio signal. Both function block A and function block B operate on the basis of a predefined sample block size. Function block B operates on the basis of a smaller sample block size than function block A. For example, the operation of function block A can be based on a block size of 1024 samples, while the operation of function block B is based on a block size of 64 samples. Function block B therefore influences the gain factor stored in memory area 68 and read out by function block C for amplification purposes in short cycles of only 64 samples.

[0057] The control speed of function block B, which depends in particular on the magnitude of the influence value d, is extremely short. In contrast, the control speed of function block A, which is essentially defined by the time constant supplied to the multiplier 64, is significantly longer. Furthermore, several control cycles of function block B can occur within a single control cycle of function block A, thus enabling a rapid response to unwanted loudness jumps. This is in Fig. 3 is illustrated by the fact that the gain factor g' read from memory area 68 can deviate by a multiple of the influence value d from the gain factor g last written to memory area 68 via the AGC 36.

[0058] In many applications, it is desirable for the control behavior of the loudness controller 22 to be neutral with respect to the sample block sizes used in function blocks A and B. However, different sample block sizes are advantageous for efficient signal processing. For example, the CPU resources required to process a single block of 1024 samples are lower than when processing 16 blocks of 64 samples each, even though the same number of samples are ultimately processed, due to reduced processing overhead and potentially more efficient implementations of the processing functions. In the exemplary embodiment, the comparatively large sample block size for function block A enables more efficient computation than using a smaller block size of, for example, 64 samples.In the present real-time system, a large block size is critical because processing the input signal, and thus reacting to changes in it, is only possible after a complete block of samples has arrived. In other words, there can be a delay of essentially the length of a sample block before an input signal peak reaches the AGC 36 for processing, allowing it to then change its gain factor. If the AGC 36's control is to start reliably as soon as the signal peak arrives at the multipliers 52, a delay in the retarders 54 corresponding to the block size is required. If the AGC 36's control is to be completed by this time, the delay value would have to be increased accordingly.

[0059] This limitation is circumvented by using different block sizes in the processing, as the operation of functional block B is based on a significantly smaller sample block size. The small sample block size in functional block B allows for a fast response to outliers in the input signal, since the corresponding processing buffers are filled more quickly, thus enabling faster signal peak processing. The delay in block C therefore only needs to account for filling a sample block buffer for block B and the corresponding control time. As a result, functional block A achieves efficient calculation while maintaining high control speed in functional block B and comparatively short delay values ​​in block C. This is advantageous for real-time processing, as the delay from the arrival of the audio signal to its output at the corresponding circuit can be minimized.

[0060] Fig. Figure 4 illustrates the structure of functional block B in detail, again for the case of stereo processing of the audio signal. The functionalities of converter 40, multiplier 42, and prefilter 44 are not discussed further, as they are already related to Fig. 2 were explained. The following description focuses on the construction and operation of the detector 46 for short-term loudness and the calculation unit 48.

[0061] In the present embodiment, the short-term loudness detector 46 essentially comprises a loudness detector 46A, which initially includes a unit 70 for each channel for calculating a mean-square value. The averaging time in this mean-square calculation can be shorter than the averaging time of the corresponding loudness calculation in the first functional block A. Thus, one can speak of determining a "short-term loudness." Upstream of the unit 46A are the previously described pre-filters 44, whose function can correspond to the definitions contained in the ITU-R standard. However, a deviation in the design of the pre-filters 44 from the ITU-R standard, for example to reduce the influence of low-frequency signal components in the short-term loudness calculation, is also possible.

[0062] The output signals of the processing units 70 are fed to an adder 72. The output signal of the adder 72, in turn, serves as the input signal of another processing unit 74. The output signal of the processing unit 74 – and thus the output signal of the loudness detector 46A – represents the resulting short-term loudness of the audio signal scaled by the multipliers 42 and is fed to the processing unit 46B.

[0063] In summary, the cascade of components 44 and 46A corresponds to a modified calculation of ITU loudness with regard to the design of the pre-filters 44 and the averaging time in the mean-square calculation 70.

[0064] In the calculation unit 46B, the difference between the loudness of the scaled audio signal and a loudness limit is determined using an adder 76. The loudness limit is the additive combination of the target loudness and a tolerance value and is calculated using another adder 78. The target loudness fed to adder 78 is the same target loudness that is also fed to adder 60 according to Fig. 3 is supplied. The tolerance value can be a fixed or variable value.

[0065] The choice of tolerance value generally depends on the measurement method used to determine short-term loudness and the system specifications regarding tolerable loudness jumps at the system output. It should be noted that a very small tolerance value can lead to a reduction in the overall loudness of the audio signal, since any brief exceedance of the loudness limit results in a reduction of the AGC 36 gain factor. In a variant using a short-term measurement based on the described modified ITU loudness with a time constant of 5 ms, the tolerance value can, for example, be in the range of 6–15 LKFS above the AGC 36 target value.

[0066] The difference between the calculated loudness threshold and the loudness of the scaled audio signal is fed to a function block 80 of the processing unit 48. A value y (for example, from an assigned memory area) is also fed to function block 80. If function block 80 determines that the loudness of the scaled audio signal exceeds the loudness threshold, the influence value d, output by function block 80, is set to the value of y. Otherwise, the influence value d is set to zero. An adder 82 downstream of function block 80 adds the output value of function block 80 to the gain factor currently stored in memory area 68. The result of this calculation is written back to memory area 68.The inherent time delay from the start of processing in function block B until the update of the value retrievable from memory area 68 from one processing cycle is represented in the block diagram by the delay element 84.

[0067] The operating cycle of function block B described above can occur multiple times during an operating cycle of function block A. This is because, as explained above, function block B operates on the basis of a smaller sample block size than function block A. Furthermore, it should be noted that, in addition to the smaller sample block size chosen for function block B, the magnitude of the influence value d significantly affects the rate of loudness reduction (and thus the control speed or duration) attributable to function block B. In any case, the rate of loudness reduction to the loudness threshold attributable to function block B is considerably higher than the rate attributable to function block A in order to reduce the loudness of the audio signal to the target loudness.

[0068] Fig. Figure 5 uses a flowchart to illustrate the functionality of function block B. The in Fig. The 5 steps shown are optionally started for each sample or each sample block (with 64 samples) in step 502.

[0069] In step 504, the current gain factor ("AGC gain") is read from memory area 68. Subsequently, in step 506, the input audio signal is scaled by the gain factor read from memory area 68 (using converter 40 and multiplier 42 according to...). Fig. 4).

[0070] For the scaled audio signal, in a further step 508, which is in detector 46B in Fig. In step 4, the short-term loudness is calculated as described above. In a subsequent analysis step 510, it is checked whether the short-term loudness determined in this way exceeds the loudness limit. As already explained, the loudness limit is the additive combination of the target loudness of AGC 36 and a tolerance value.

[0071] Should the short-term loudness actually exceed the loudness threshold, the gain factor is reduced in step 512 based on the influence value d=y. This reduction involves a write access to memory area 68.

[0072] If, however, it is determined in step 510 that the loudness limit is not exceeded, the procedure ends in step 514. Consequently, function block B is not activated further (corresponding to the case d=0 in function block 80 according to...). Fig. 4).

[0073] The following section discusses the influence of loudness control 22 on an audio signal using several diagrams. Fig. 6 and Fig. Figures 7 show, at the top and bottom respectively, the time course of an input and output audio signal of the loudness control 22. In the middle, the time course of the gain factor (in dB) stored in memory area 68 for scaling the input audio signal is shown.

[0074] The in Fig. The input audio signal shown above exhibits a rising signal level. The increase at time 1:00 min is relatively small, while the increase at time 2:00 min is relatively large.

[0075] As in Fig. As shown in Figure 6, the loudness of the input signal is below the target loudness until time 1:00 min. Therefore, the gain factor is slowly increased to approximately 4 dB by function block A (according to the release time factor). Function block B is not active.

[0076] Due to the loudness jump at 1:00 min, the loudness of the input signal is at the target loudness level. The applied gain factor to achieve the target loudness must therefore be reduced from the previously applied approximately 4 dB to 0 dB. During the increase in the input signal at 1:00 min and up to 2:00 min, the short-term loudness of the scaled signal (not shown) measured in block B does not exceed the set loudness threshold. For this reason, function block A slowly reduces the gain factor (according to the attack time factor), while function block B remains inactive.

[0077] At 2:00 minutes, a sharp loudness jump occurs, causing both the target loudness and the loudness threshold to be exceeded. This exceedance of the loudness threshold triggers an extremely rapid reduction in the gain factor by function block B. This rapid reduction continues until the resulting short-term loudness of the scaled audio signal no longer exceeds the loudness threshold. Subsequently, a gradual adjustment back to the target loudness is performed solely by function block A.

[0078] As in the output signal of the volume control in Fig. As shown in Figure 6 below, due to the intervention of function block B at time 2:00 min, there is no excessive loudness jump in the output signal, although there is a strong loudness jump on the input side.

[0079] Fig. Figure 7 illustrates the operation of the loudness control 22 in a similar way to that shown in Fig. 6, but for a different signal waveform of the input audio signal. In Fig. Figure 7 illustrates in particular that the reduction in gain by function block B is less than the required reduction in the gain factor to achieve the target loudness. This fact is also related to the fact that the loudness threshold used by function block B is one tolerance value higher than the target loudness used by function block A.

[0080] Fig. Figure 8 shows a close-up of the upper diagram from Fig. 7 (top) and the middle diagram from Fig. 7 (below) between the times 2:32 min and 2:33 min. The second diagram from the top in Fig. Figure 8 shows the result of the loudness measurement in function block B (in LKFS), and the third diagram from the top shows the duration during which function block B accesses memory area 68 to repeatedly reduce the currently valid gain factor by the influence value d=y.

[0081] The fact that the loudness measured in function block B is above the loudness limit (target loudness plus tolerance value) during the control period is clearly visible, but approaches the loudness limit very quickly. This is particularly illustrated by... Fig. 8, that the rate at which the loudness of the audio signal is reduced by function block B is significantly higher than the rate at which function block A reduces the loudness of the audio signal to the target loudness by raising or lowering it.

[0082] While the Fig. Figures 6 to 8 illustrate the function of the loudness control 22 for a test signal. Fig. Figure 9 illustrates the operation of the loudness control 22 for a sequence of three musical pieces with increasing loudness. The first piece lasts from 0 to approximately 60 seconds, the second from approximately 60 to approximately 100 seconds, followed by a pause of approximately 10 seconds, and the third from approximately 110 seconds until the end of the diagram. Again, the upper diagram shows the input audio signal and the lower diagram the output audio signal of the loudness control 22. The middle diagram shows the curve of the gain factor and the corresponding scaling when both the control in function block A and the control in function block B are active. Fig. Figure 10, however, illustrates the control behavior of the volume control 22 when function block B remains switched off (or is missing) with otherwise identical settings to the recording in Fig. 9.

[0083] The following can be clearly seen in the lower diagram: Fig. 10. A significant loudness jump in the output signal occurs during the transition from the first to the second track (at approximately 60 seconds), followed by the adjustment of the AGC 36, which only very slowly approximates the loudness of the subsequent signal to the target loudness. Furthermore, it is evident that the second track, in particular, is actually amplified in loudness over large portions by the AGC 36, even though the set gain factor results in a signal loudness above the target loudness (recognizable by the continuous decrease in the gain factor following the transition from the first to the second track).

[0084] In contrast, the very quick regulation at the same location in Fig. 9. It is evident that by intervening in function block B, a largely correct scaling can be set as early as the beginning of the second track, so that the subsequent adjustment by AGC 36 is significantly less than in Fig. 10. The one in Fig. 10. A noticeable large loudness jump during the transition from the first to the second track can also be prevented by the intervention of function block B.

[0085] It is also evident in Fig.9. Within the first track, a significant reduction in the gain factor of the AGC 36 occurs only upon the arrival of the first signal peak, so that the signal's dynamics in the 10-30 s range are largely preserved, similar to the variant without functional block B. The reduction in the gain factor at approximately 10 s also occurs directly before the arrival of a transient in the input signal, thus in a psychoacoustically advantageous manner. In the quieter passage of the first track (30-60 s), the gain factor is then gradually increased based on the AGC release time to achieve a slow adjustment of the audio signal's loudness while keeping the changes barely audible.

[0086] According to one embodiment, parts of the processing in function block D can be deactivated when silence is detected. This avoids unnecessary operation of function block B (and thus necessary calculations), and, more importantly, prevents incorrect control by the AGC 36. This is necessary to prevent the AGC 36 from increasing the applied gain factor during silent passages to raise the signal to the target loudness, which would ultimately amplify the system's background noise. Furthermore, it would be conceivable to deactivate the control when, during radio operation, the radio receiver 14 detects poor signal quality and therefore reduces the volume of the audio signal, thus counteracting any increase of the deliberately attenuated signal to the target loudness.For this purpose, the loudness control 22 can have a separate interface to the radio receiver 14 in order to obtain information regarding the signal quality from the radio receiver 14.

[0087] As the description of the implementation examples shows, the loudness adjustment technique presented here enables efficient adaptation of loudness to, for example, source and track changes. The resulting psychoacoustic advantages can, in many implementations, eliminate the need for user readjustment of the loudness.

Claims

[1] A method (500) for adjusting the loudness of an audio signal, comprising the steps of: Analyzing (504, 506, 508) a loudness of the audio signal with respect to a target loudness and determining a gain factor for increasing or decreasing the loudness of the audio signal in order to continuously increase the loudness of the audio signal to the target loudness at a first rate or to decrease it to the target loudness at a second rate; analyzing (510) a resulting loudness of the audio signal scaled on the basis of the determined gain factor with respect to exceeding a loudness threshold; and if the resulting loudness exceeds the loudness threshold, lowering (512) the loudness of the audio signal at a second rate that is higher than both the first first rate and the second first rate. [2] The method of claim 1, wherein lowering the loudness of the audio signal at the second rate brings the audio signal to the loudness threshold. [3] A method according to claim 1 or 2, wherein the lowering of the loudness of the audio signal at the second rate occurs continuously. [4] Method according to one of the preceding claims, wherein in order to reduce the loudness of the audio signal at the second rate, the determined gain factor is modified and the audio signal is scaled with the modified gain factor. [5] Method according to one of the preceding claims, wherein the loudness threshold is derived from the target loudness. [6] A method according to claim 5, wherein the loudness threshold is a tolerance value above the target loudness. [7] A method according to any one of the preceding claims, wherein the first loudness increase rate or the second loudness decrease rate is in a range between approximately 0.1 dB / s and 2 dB / s and / or the second loudness decrease rate is in a range between approximately 0.1 dB / ms and 2 dB / ms. [8] A method according to any one of the preceding claims, wherein the first loudness increase rate or the second loudness decrease rate results in a control period of approximately 5 s to 5 min and / or the second loudness decrease rate results in a control period of approximately 1 ms to 1 s. [9] Method according to one of the preceding claims, wherein the first installment and the second installment are different, and the increase in loudness with the first first rate is slower than the decrease in loudness with the second first rate. [10] Method according to one of the preceding claims, wherein the analysis of the loudness of the audio signal with respect to the target loudness and the determination of the gain factor are carried out in conjunction with an automatic gain control. [11] A method according to claim 8, wherein the automatic gain control is of the feed forward type. [12] A method according to any one of the preceding claims, wherein analyzing the resulting loudness of the audio signal scaled with the determined gain factor comprises signal peak detection. [13] A method according to any preceding claim, further comprising delaying the audio signal to synchronize with the lowering of the loudness of the audio signal at the second rate. [14] Method according to one of the preceding claims, further comprising performing a silence detection for the audio signal, wherein if silence is detected, at least one of the following steps is omitted: - an adjustment of the gain factor with the first first rate or the second first rate to achieve the target loudness - scaling the audio signal; - analyzing the resulting loudness of the audio signal scaled with the determined gain factor; - lowering the loudness of the audio signal at the second rate. [15] Method according to one of the preceding claims, wherein - the audio signal is based on a radio signal and in case of poor radio reception at least one of the following steps is omitted: - an adjustment of the gain factor with the first first rate or the second first rate to achieve the target loudness - scaling the audio signal; - analyzing the resulting loudness of the audio signal scaled with the determined gain factor; - lowering the loudness of the audio signal at the second rate. [16] A method according to any one of the preceding claims, wherein the audio signal is sampled and the analysis of the loudness of the scaled audio signal is based on a smaller sample block size than the analysis of the loudness of the non-scaled audio signal. [17] A device (22) for adjusting the loudness of an audio signal, the device comprising processor hardware and software, the processor hardware being configured to perform a method according to any one of the preceding claims under the control of the software. [18] A motor vehicle audio system (10) comprising the device (22) according to claim 17.

Citation Information

Patent Citations

  • Device for use in radio of motor car for controlling volume of different audio signals of e.g. TV transmitter, has memory unit with volume control information used for controlling volume of selected audio signal during selection of signal

    DE102010003306A1

  • Systems and methods for controlling audio loudness

    US20080253586A1

  • Loudness consistency at program boundaries

    US20100272290A1