WO2006075264A1

WO2006075264A1 - Signal processing arrangement and audio system for and method of frequency-dependent amplifying of the sound level of audio signals

Info

Publication number: WO2006075264A1
Application number: PCT/IB2006/050040
Authority: WO
Inventors: Paul Ullmann
Original assignee: Koninklijke Philips Electronics NV
Current assignee: Koninklijke Philips NV
Priority date: 2005-01-14
Filing date: 2006-01-05
Publication date: 2006-07-20
Anticipated expiration: 2007-07-14
Also published as: JP2008527882A; CN101103525A

Abstract

The invention relates to a method for the frequency-dependent amplifying of the sound level of audio signals, in which an audio signal (Sig(f)) is divided, at least in a certain frequency range, into a number (n) of frequency bands (f1...f10), a particular auditory threshold (Th(f1)...Th(f10)) is specified for each frequency band (f1 ...f10), for each frequency band (f1...f10) the sound level is compared with the respective pre-determined auditory threshold (Th(f1)...Th(f10)), and the audio signal (Sig(f2), Sig(f3), Sig(f4)) is amplified at least in those frequency bands (f2, f3, f4) in which the sound level lies just below the respective auditory threshold (Th(f2), Th(f3), Th(f4)). The invention furthermore relates a signal processing arrangement for implementing the method, as well as an audio system with such a signal processing arrangement.

Description

Signal processing arrangement and audio system for and method of frequency-dependent amplifying of the sound level of audio signals

FIELD OF THE INVENTION

The invention relates to a method of frequency-dependent amplification of the sound level of audio signals.

The invention furthermore relates to a signal processing arrangement for implementing such a method.

The invention furthermore relates to an audio system with such a signal processing arrangement.

BACKGROUND OF THE INVENTION The auditory threshold in humans is known to be a function of frequency, with the range of greatest sensitivity for the human ear lying at a frequency of approximately 300 Hz to 3 kHz. At low frequencies, the auditory threshold is relatively high (a high sound pressure is necessary), and at higher frequencies in turn it shows a slight rise - until abruptly the upper audibility limit is reached, which in younger people lies at around 18 kHz and declines with increasing age.

It is furthermore known that perception of harmonics is decisive for authentic perception of a sound or of the instrument that produces the sound.

An evident phenomenon is that in the reproduction of sound events, such as for example music of low sound levels, components of the acoustic pattern, in other words certain harmonics, are weakened below the auditory threshold. A listener experiences the acoustic pattern as having "something missing", the music loses some of its atmosphere, and the sound seems dull. These factors produce the impression of hearing the music "too quiet".

With regard to the reproduction of music, there is furthermore only one truly correct sound level - namely, the sound level that corresponds to the actual event, for example the recording of the music. However, since in reproduction situations one often intentionally selects a lower volume than the correct one (for example to avoid disturbing others), one makes a compromise in respect of authenticity (to say in respect of atmosphere). In order to remedy this restriction 1 in many audio and audio/video devices the so-called "loudness" function is employed. This loudness function represents a kind of preselection on an equalizer, in which the sounds lying at low frequencies are amplified strongly, the high frequency sounds are amplified somewhat, and the sounds in a middle frequency range are not amplified at all. In this way, the low and high-frequency portions are audible even at low volume. A circuit for volume control corresponding to this loudness function is described for example in Zastrow, Peter: "Phonotechnik", 1^st edition, Frankfurter Fachverlag, 1979, p. 132 - 139.

This loudness function fundamentally already yields an improvement in the acoustic pattern, but is far from capable of creating a perfect or near-perfect acoustic pattern for the user. The problem lies in the fact that the sound level of the high and lower sounds are respectively altered by a fixed, pre-determined amount, through which in turn no optimum authenticity of the acoustic pattern is achieved.

OBJECT AND SUMMARY OF THE INVENTION

It is an object of the invention to create a method in accordance with the type stated in the first paragraph, a signal processing arrangement in accordance with the type stated in the second paragraph, and an audio system in accordance with the type stated in the third paragraph, in which a considerably better modification of the acoustic pattern is achieved, with which a perfect or near-perfect, in other words a considerably more authentic acoustic pattern, can be achieved. To achieve the object mentioned above, in the case of a method according to the invention, features according to the invention are provided such that a method according to the invention can be characterized in the manner stated below, namely:

Method for frequency-dependent amplifying of the sound level of an audio signal, in which method the audio signal is divided, at least in a certain frequency range, into a number of frequency bands, and a particular auditory threshold is specified for each frequency band, and for each frequency band the sound level of the audio signal in the frequency band is compared with the respective specified auditory threshold, and the audio signal in those frequency bands in which the sound level lies just below the respective auditory threshold is amplified in such a way that the sound level of the amplified signal in these frequency bands lies above the auditory threshold.

To achieve the object mentioned above, a signal processing arrangement according to the invention has means for implementing such a method.

Furthermore, to achieve the object mentioned above, an audio system according to the invention has a signal processing arrangement, as mentioned above. Through the features according to the invention, the advantage is achieved that in the case of amplification of audio signals as per the invention, a frequency-dependent auditory threshold of a user is taken into account, and a selective amplification of the audio signals takes place in particular frequency ranges that depend on the respective auditory threshold, through which the user is provided with a considerably better acoustic pattern than in the case of the familiar loudness function described above, in which amplification takes place only in the fixed, pre-determined frequency ranges.

In principle, the audio signal may be split into frequency bands only in one particular frequency range or in several frequency ranges of the audible spectrum, and amplification takes place if necessary. But for an optimally amplified audio signal, it is advantageous if essentially the overall audible range of the spectrum is split into corresponding frequency bands and corresponding amplification takes place.

At this point let it furthermore be pointed out that for the concept "sound level in a frequency band", alternatively one can also use the concept of spectral energy within the frequency band.

In principle, for the purposes of the invention it is sufficient if an audio signal is then amplified above the auditory threshold when it lies "just" below the respective auditory threshold. Fundamentally it is of course also possible likewise to amplify audio signals that lie further below the respective auditory threshold. If this is the case, care must however be taken that these are not amplified above the auditory threshold, since otherwise the authenticity of the acoustic pattern can suffer.

For preference, however, in order to achieve a particularly authentic acoustic pattern, the feature of claim 2 are furthermore provided.

It is particularly advantageous if, in the case of the invention, the measures of claim 3 are realized. Through these measures, considerably optimized amplification can take place, since for the various frequency bands, individual amplification takes place, corresponding to the respective sound level of the audio signal and taking into account the auditory threshold.

In particular, it is advantageous if the measures of claim 4 are provided. Through these measures, once again a considerably more precise amplification is made possible, since for example in this way, one is better able to take account of differences in sensitivity in the ears of a user.

In order that there is no amplification of those frequency ranges of an audio signal, in particular of each channel of the audio signal, which are well audible in any case, the measures of claim 5 are furthermore provided.

In order to obtain as authentic an acoustic pattern as possible, it is necessary that the amplification in the various frequency ranges makes the best possible allowance for the relations between the sound level of this frequency band, the auditory threshold and the threshold value. This taken into account in the optimum manner by the measures of claim 6.

In this connection, the measures of claim 7 are furthermore provided.

In order not to alter the sound level of the overall audio signal (per channel) too heavily, the measures of claim 8 and in particular of claim 9 are furthermore provided, through which, whilst the overall characteristic continues to correspond to the "new" amplified acoustic pattern of the audio signal, through the even reduction the overall sound level of the "new" audio signal is reduced. In particular, through the measures of claim 9 a standardization of the "new" audio signal to approximately the original sound level takes place, through which the sound levels of the original and of the new, modified audio signal hardly differ, or do not differ at all. These measures are based on the fact that in the normal case, the overall sound level of the audio signal (for the channel in question) essentially results from spectral portions that lie clearly above the auditory threshold. On the other hand, amplification of the signals in the border areas does not make any significant contribution to the overall sound level. This means that for this "normal" case, the overall amplification described above need only be reduced to an insignificant degree.

If on the other hand the whole spectrum was too quiet prior to processing, and that therefore there are no dominant spectral ranges (for example silence, quiet background noise, applause etc.), it can happen that amplification is assigned to all the bands, and this is then essentially reversed in the standardization step. However, this is right and desirable for hearing.

The choice of the width of the frequency bands can be respectively adapted for the specific case, and various possibilities for this are shown in the description. It is however particularly advantageous if the measures of claim 10 are included, through which particularly good differential treatment of adjacent notes of different volumes becomes possible.

In order to achieve the highest possible level of authenticity, the measures of claim 11 are furthermore provided. According to these measures, the audio signal is amplified only when the sound event shows a lower sound level than the original sound level of the original sound event, whilst amplification is not necessary in the case of the original sound level or above.

In principle, for the method according to this invention, a typical auditory threshold for a user can be specified. However, the method according to this invention can be adapted particularly well and individually to a user if the measures of claim 12 are provided. As has already been mentioned above, it is advantageous if each channel of the audio signal is amplified separately. For that reason it is also particularly advantageous if furthermore the measures of claim 13 are provided, since then besides each channel of the audio signal, the respective relevant individual auditory threshold of the user is also taken into account for this channel. Here, the auditory threshold can be best established in that finally the measures of claim 14 are provided.

For an audio system as mentioned at the outset, it is furthermore advantageous if the measures of claim 17 are realized. With an audio system of this type, the overall invention can be realized in a compact manner in a single system, and no separate components are necessary.

These and other aspects of the invention are apparent from and will be elucidated with reference to the embodiments described hereinafter.

BRIEF DESCRIPTION OF THE DRAWINGS In the following, the invention is described on the basis of the design examples shown in the Figures, which are however non-limitative.

Figure 1 shows, schematically, the progression of the auditory threshold as a function of frequency for a young person with normal hearing.

Figure 2 shows typical progressions of auditory thresholds for different people as a function of frequency in a range of 1 kHz - 6 kHz.

Figure 3 shows, schematically, the relations between auditory threshold, signal level (volume) and the resulting threshold value for a channel of an audio signal, on the basis of a simplified example.

Figure 4 shows an example progression of the additional amplification of an audio signal according to the invention.

Figure 5 shows an example of the progression of the maximum permitted amplification of an audio signal as a function of the current sound level of the audio signal.

Figure 6 shows a schematic block diagram of an audio system according to the invention. DESCRIPTION OF EMBODIMENTS

Figure 1 shows schematically the progression of the auditory threshold as a function of frequency for a young person with normal hearing, as has already been discussed in the introduction. Sound pressure p[dB] is shown as a function of frequency f [Hz].

In the case of the familiar loudness function, as already described in the introduction, this individual progression of the auditory threshold is not taken into account. To sum up, it can be stated that the conventional loudness function does not take into account: a) the actual auditory threshold of a user, in particular not this auditory threshold as a function of frequency (see for example Figure 1); b) differences in the auditory threshold between the right and left ears of a user; c) the current absolute sound pressure to which each of the user's ears is exposed; d) the different spectral components of the sound.

Re a) The auditory threshold varies with time, in other words with the age of the user, as well as between the left and right ear. Other influences that affect the auditory threshold are genetic predispositions and/or noise that acts on a user over a longer period of time. However, no general statements can be made, particularly not concerning the point in time at which the ears will begin to lose sensitivity. Figure 2 shows typical progressions of auditory thresholds for different people (Mx: M = male, x = age; Wx: W = female, x = age) as a function of frequency in a range of 1 kHz - 6 kHz.

As a rule, sensitivity in the ears starts to decrease earlier in men than in women, which can be explained by, amongst other things, the fact that men are more often employed at noisy workplaces. However, as already indicated, the range of dispersion of the individual values is very high. Re b) Similar considerations apply for the differences between a user's left and right ears. In the case of young, healthy people, both ears are of a similar sensitivity, but differences between the ears can develop over time. The auditory thresholds for the left and right ears can differ by up to 10 dB - 20 dB in a middle-aged person.

Re c) If in Figure 2, one looks at the auditory threshold for an average 60- year-old man in the case of a 4 kHz sound, the ears are exposed to an average sound pressure p of approximately 40 dB. If this person is located a little further away from the sound source, and the sound pressure level is for example only 30 dB, then this person will no longer hear the sound. For a correction that is as true to reality as possible, it would be necessary to carry out the sound correction on the basis of the absolute sound pressure, which in turn depends on the source signal, possibly the bass and treble amplification, the settings of the sound processor - for example for Surround-Sound -, as well as on the loudspeaker efficiency, the distance of the loudspeakers from the listener, the acoustic situation in the room (reflections increase the SPL), etc. Furthermore, one must for example also take into account whether the listener is using headphones.

In the case of the familiar loudness functions however, this is in no way taken into account, since here the amplification is undertaken only on a relative basis, that is to say only depending on the sound level of the sound event. Re d) The familiar loudness function amplifies the sound in particular frequency ranges. This ignores the fact that the sound can comprise many components with the most varied intensities. If one assumes for example that one note occurs at 4 kHz together with a weak note at 5 kHz, then in the case of the loudness function, both notes are amplified by a certain fixed amount, although the 4 kHz note would have been easily audible even without amplification. This is unfavorable, since such amplification seems unnatural and strange.

With this invention, the disadvantages described above can now be largely or overallly avoided, in that in the case of the amplification of an audio signal, the individual auditory threshold of a person is taken into account, and a selective amplification of the audio signals takes place in particular frequency ranges that depend on the respective auditory threshold, thus resulting in a considerably better acoustic pattern for the user than in the case of the familiar loudness function described above, where amplification takes place only in fixed, predetermined frequency ranges.

It is particularly advantageous if each channel of an audio system SYS (see Figure 6) is taken into account, since through these measures once again a considerably more precise amplification is made possible, since for example in this way it is also possible to take better account of differences in sensitivity between the ears of a user.

It is pointed out at this point that the term "audio system" is to be understood as meaning any system or device for reproducing audio signals, in particular also audio/video systems (A/V systems). Such an audio system SYS is shown in more detail, by way of an example, in Figure 6.

The method according to this invention is explained in greater detail below, on the basis of Figures 3 to 5. For the Figures 3 to 5 and the more detailed description below, the definitions given now low shall be used. All variables are in decibels (dB).

Th(f) ... individual auditory threshold of a person with headphones or loudspeakers, in relation to a particular channel (left, right, surround, ...) for a particular frequency f (or for a particular frequency band around the frequency f), in relation to the (random) dB scale of the audio or audio-video system, wherein the effects of the volume control are also taken into account.

Sig(f) ... signal level (spectral energy) of the audio signal (for this particular channel) after volume control at the frequency f or in a frequency band around the frequency f on the dB scale of the audio or audio-video system.

MA ... maximum permitted amplification in dB (for example amplification "Off, 10 dB, 20 dB, 30 dB).

ML ... threshold value; the resulting threshold value Th(f) + ML represents the maximum signal strength of the audio signal Sig(f) that is still amplified (this variable defines the so-called "fade-out point", at which the additional amplification becomes zero).

Amp(f) ... additional (uncorrected) amplification of the audio signal (for one channel) in the frequency band defined by the frequency f.

AmpC(f) ... additional, corrected amplification of the audio signal (for one channel) at the frequency f or in the frequency band defined by the frequency f .

SE ... overall spectral energy (corresponds to the overall sound level of the audio signal, here for one channel) that is perceived by a user without amplification according to this invention, averaged over all frequency bands.

SEU ... overall spectral energy per frequency band that is perceived by a user with uncorrected amplification according to this invention, averaged over all frequency bands.

Fundamentally the invention is based on the fact that for the frequency- dependent amplification of the sound level of an audio signal Sig(f), this audio signal Sig(f) is divided into a number n of frequency bands fl ...flO, and for each frequency band fl ...f 10 a particular auditory threshold Th(f) is specified. Furthermore, for each frequency band fl...flθ, the sound level is compared with the respective specified auditory threshold Th(fl)...Th(flO), and the audio signal is amplified in those frequency bands in which the sound level lies close to the respective auditory threshold.

Figure 3 now shows the progression of the audio signal Sig(f) as a function of frequency f for the illustrated channel of the audio system, that is to say the current spectrum of this channel Sig(f) without amplification as a function of frequency f. Here, the frequency f is entered in a range between 20 Hz and 20 kHz on a logarithmic scale. Entered on the y axis is the signal level or the spectral energy (volume per frequency unit). Figure 3 also shows the individual auditory threshold Th(f) of a user as a function of frequency f for this channel. In accordance with the invention, the frequency range is now divided into a particular number n of (narrow) frequency bands fl...flθ, a typical value being for example n = 200.

Typically, in the simplest case the frequency bands on a logarithmic scale (seen logarithmically) are equally wide, since this comes close to the physiology of the ear, that is to say f(band end) = c * f(band start), wherein the factor c remains constant over the whole auditory range. If it were the case that c=2, then (inexpediently) in each case one octave would be combined into one band, with c=1.0595... (12^th root of 2) one band respectively corresponds respectively to one semi-tone step, which does represent a sensible resolution.

Alternatively, it can be provided that the frequency bands are kept denser in those frequency ranges in which the human ear reacts most sensitively, in other words between approximately 200 Hz and 3 kHz. Above 3 kHz and below 200 Hz, the frequency bands can, for expedience, be selected somewhat broader.

In principle, a decomposition could be provided into such narrow frequency bands that de facto a frequency band consists only of one concrete frequency. This situation, in other words that in the case of quite concrete frequencies a comparison of Sig(f) and of the auditory threshold Th(f) takes place, and possibly an amplification, is also included in the formulation "frequency band". However, taking into account the available calculation capacity as well as the restricted frequency resolution capacity of the human ear, it is expedient if a "frequency band" actually comprises a range of frequencies and not just one frequency. In the further description and the formulae given, for the purpose of clarity we shall assume frequency bands of equal width (on a logarithmic scale). Further, in Figure 3 only n = 10 frequency bands fl...flθ are assumed, in order to allow greater clarity of representation.

The division of the audio signal Sig(f) into frequency ranges fl...flθ takes place either as mentioned above in the digital signal processor DSP of the audio system SYS (see Figure 6) or, if for example we are dealing with compressed audio signals, these are already present in a corresponding form.

As already mentioned, Figure 3 furthermore shows a continuous progression of the individual auditory threshold Th(f) - usually however this progression is already present in the quantized form Th(fl) ... Th(flO), which is likewise shown, which already results for example through determination of the auditory threshold for a user for one channel, as is explained in greater detail below.

In the individual frequency bands, it is now checked whether the audio signal Sig(f) is close to the respective auditory threshold Th(f), and if necessary the audio signal Sig(f) is amplified in the corresponding frequency band, as shown in Figure 3 by way of an example for the frequency ranges f2, f3, f4.

The term "close" in relation to the auditory threshold Th(f) is to be understood such that essentially an amplification takes place in those frequency ranges f where the signal Sig(f) lies below the auditory threshold Th(f) by at most a value MA for the maximum amplification. Preferably, in particular in order to obtain an authentic acoustic pattern, an amplification is also provided of those signals Sig(f) whose value lies directly on the auditory threshold Th(f) or "close" above it. Upwardly, the window defined by the term "close" can likewise be limited by the value MA, so that in principle the term "close" can be characterized in that the value of Sig(f) lies roughly between Th(f) - MA and Th(f) + MA.

As described in greater detail below, the upper value for "close" can be defined still better by the threshold value ML, as is explained in greater detail below.

It is mentioned at this point that in principle it is naturally also possible to amplify such values of Sig(f) as lie below the auditory threshold Th(f) by more than the maximum amplification MA, wherein however this has no effect on the resulting acoustic pattern, since in these frequency ranges the sound level is not raised beyond the auditory threshold in this range through such amplification.

It is mentioned at this point that usually this comparison is carried out for all frequency bands, and if necessary a corresponding amplification is carried out. In principle it can however be provided that a corresponding comparison is carried out only in one particular frequency range or in different frequency ranges, and in other frequency ranges a comparison or amplification is not realized.

For the calculation of the amplification Amp(f), a variable is provided - in this case the threshold value ML. If the signal level Sig(f) is equal to or more than ML over the auditory threshold, there is no amplification for this signal band. Shown in Figure 3 is the quantized progression of the resulting threshold value Th(f) + ML, above which no further amplification of the audio signal Sig(f) takes place.

The additional uncorrected amplification Amp(f) for a frequency band around the frequency f can now be stated for a concrete embodiment as Amp(f) = (Th(f) + ML - Sig(f)) * MA/(ML+MA) for Sig(f) < Th(f) +ML, and

Amp(f) = 0 for Sig(f) > Th(f) +ML. The progression of the uncorrected amplified signal is shown on the logarithmic scale of Figure 3 as Sig(f) + Amp(f).

The progression of the uncorrected amplification Amp(f) before the concluding standardization step described below is shown in Figure 4 as variant α. As one can see for this variant α, the amplification lies, depending on the value for Sig(f) at the auditory threshold Th(f), at the value MA, and at the resulting threshold value Th(f) + ML (as well as for values of Sig(f) over this resulting threshold value) at zero.

Towards smaller signal levels of Sig(f), the progression of Amp(f) shows a continuous rise. However, contrary to the illustration in Figure 4, for Sig(f) < Th(f) - MA usually the value for Amp(f) is set to zero, or the gradient of Amp(f) is to be selected such that for the amplified signal Sig(f) + Amp(f), the sound level does not lie above the auditory threshold Th(f).

The parameter ML here is defined via the spectral energy of the channel as ML = cl*SE + c2, with the two constant cl and c2. Preferably, these constants assume the value cl = 1 and c2 = 0. The parameter ML is thus a function of the perceived spectral energy of the original signal, that is to say of the signal without amplification according to the invention.

Through the dependence of the parameter ML on the spectral energy, it is achieved that an audio signal which for example is located close to or below the auditory threshold in the case of all frequencies, is not amplified, so that silent or quiet passages in a piece of music, or a break between two pieces of music, are not amplified or only weakly amplified.

The overall spectral energy SE of the audio signal for the channel shown, and thus the resulting sound level of this channel, is the difference between Sig(f) and Th(f), totaled up over all frequency bands SE = ∑_x=1..._n {[Sig(f_x) - Th(f_x)]/n} for Sig(f_x) > Th(f_x), and SE = O for Sig(f_x) < Th(f_x), wherein n indicates the number of those frequency bands into which the spectrum of the original signal is divided, and f_x characterizes the individual frequency bands. The value for SE is shown in Figure 3 by the hatched area.

Finally, the amplification is further standardized to the effect that the overall spectral energy that is perceived by a user is roughly the same before and after amplification, in that the amplified signal is reduced by the same value in all frequency bands.

The corrected (standardized) amplification in a frequency band that thus results then yields

AmpC(f) = Amp(f) + SE - SEU, wherein

SEU = ∑_x=1..._n {[Sig(f_x) + Amp(f_x) - Th(f_x)]/n} for Sig(f_x) + Amp(f_x) > Th(f_x), and SEU = 0 for Sig(f_x) + Amp(f_x) < Th(f_x).

In the selected logarithmic representation, the amplified, corrected signal then results as Sig(f) + AmpC(f).

With this method of proceeding, a comparatively small correction is achieved, through which a fresh calculation of the amplification - which otherwise, in the case of too strong a correction, could be reduced again too heavily - by means of an iterative calculation is no longer necessary.

With the determination, described above, of the amplification Amp(f) of an audio signal (of a channel of an audio signal) as well as of the corrected, standardized amplification AmpC(f), very good acoustic results are achieved. In principle it would be possible, in the case of a multi-channel audio system, to amplify just one or a few channels as described above. Naturally however, optimum and authentic results are achieved in the modification of an audio signal in particular when all channels are modified accordingly, and in particular for each channel the corresponding individual auditory threshold of the user is taken into account. Besides the variant α, shown in Figure 4, for the amplification Amp(f), another possible variant β is shown. In the case of this variant β, for Sig(f) = Th(f) - MA the amplification assumes the value Amp(f) = MA. Above Th(f) - MA, Amp(f) then decreases in a linear manner towards Th(f) + ML, and at Th(f) + ML (and for values above that) the amplification becomes zero. Below the threshold value Th(f) - MA, the amplification is Amp(f) = MA according to Figure 4. In this range however, for the user this constant amplification MA no longer has any audible effect on the acoustic pattern, so that usually in this range, amplification is no longer carried out, and accordingly Amp(f) = 0 applies for Sig(f) < Th(f) - MA.

Another variant, not shown here, for the selection of the characteristic curve Amp(f) is that the gradient is derived directly from the value MA, instead of making the gradient dependent on ML, as described above, and the amplification for signal values above Th(f) + ML once again becomes zero. Below that however, the Amp(f) progression is not defined by the condition that Amp(f) at the point Th(f) - MA should be equal to MA, but is defined in a fixed manner by a gradient, wherein this gradient is given by = c * MA, wherein c is any positive constant, so that the gradient never becomes steeper than -45°, even in the case of the highest possible value for MA.

A further advantageous supplementation of the invention is finally explained in greater detail on the basis of Figure 5. Until now, the concept of the "original sound level" of a sound event, for example a piece of music, has not yet been taken into account. As already mention in the introduction to the description, the listener perceives a sound or acoustic event as "natural" if it is reproduced at the "original volume" (as for example in the concert hall, etc.). Particularly advantageous therefore is an embodiment of the amplification of audio signals according to the invention, where this amplification behaves completely neutrally at the original sound level (and at sound levels above the original sound level). This is achieved as follows:

Currently, in the case of a music recording on a compact disc (CD) or similar, we have no information about the original sound level. If such information is available, of course for preference this original sound level can be used.

If this information is not automatically available, the listener must define the "original sound level" himself. In other words, before actually listening to the recording on an audio system, advantageously he will be played a loud passage (without spectral processing) and will set the volume controller such that the auditory impression appears authentic to him. The loud passage can be found by the audio system, for example in the familiar manner of a peak search or through a faster algorithm. The passage should be only relatively loud, in order to facilitate a reliable choice of the authentic sound level. It must however not be the absolutely loudest place. The value set by the user for the original sound level VoI(OL) is stored by the audio system SYS.

Figure 5 now shows the parameter MA for the maximum permitted amplification as a function of current sound level VoI which is reproduced by the audio system SYS. In the case of the original sound level VoI(OL) determined at the start, and above that, the maximum possible amplification MA becomes zero; at lower volume the maximum amplification MA increases up to a predetermined maximum value MAmax. Figure 5 shows a possible course of a curve for this rise.

A central point in the case of the present invention is the consideration of the auditory threshold Th(f) as a function of frequency f. In principle, a typical progression for the auditory threshold Th(f) could be specified. It is however much more favorable to take into account an auditory threshold Th(f) that has been determined individually for a user.

Fundamentally, the auditory threshold can be determined independently of the individual channels of the audio system SYS; although it is considerably more favorable - since clearly better acoustic results are achieved - if the auditory threshold Th(f) for a user is established for all channels of the audio system, and the amplification is carried out according to the invention separately for all channels. Here, the determination of the auditory threshold can preferably be undertaken by the user (for example at home) with the audio system SYS.

It is particularly advantageous if, for determining the frequency-dependent auditory thresholds assigned to each channel, the user is at the place where he will later prefer to listen. In this way, advantageously the effect of the room acoustics, acoustic shadowing and distance between user and loudspeaker system feed into the threshold values that are determined.

In the following, the determination of the auditory threshold is briefly explained in greater detail. Via the audio system SYS, a series of test sounds is played to the user. The user must respond to these test sounds. In the course of this, a distinction is made between all the channels of the audio system SYS. For example, for the audio system SYS shown in Figure 6, a distinction is made between the six loudspeakers SPE or, if the sounds are reproduced via the headphones HPH, a distinction is made between the left and the right speakers of the headphones HPH. In other words, the sequence of test sounds is repeated for each channel that is present.

The test series also contains sounds for checking the validity of the user responses. Such tests for determining the auditory threshold are well known in principle from the clinical field. Optionally, several data sets can be produced and called up for a user, for example to make the amplification function as per the invention available for various listening locations, in other words for different places where the user listens to music, for example, in a quality that remains constant. In this case, when activating the frequency- dependent amplification function described in this document, the user should also state the relevant listening location. Alternatively, the audio system can deduce the listening location itself, insofar as it can for example locate the remote control with which the activation is undertaken.

In each case, the measured values are stored per user. It is particularly advantageous to set up one or more data records for each user of the system. In this case, the user must also identify oneself when activating the frequency-dependent amplification function described in this document.

When determining the auditory threshold, it is not necessary to check each individual one of the frequency bands used in the algorithm. It is sufficient to record some reference points and to extrapolate the rest.

As a special form of such extrapolation, the evaluation of otoacoustic emissions can be applied. The advantage of this method is that a good frequency resolution can be achieved without any active response from the human subject being required. The method is familiar from the clinical field, and therefore requires no further explanation here. One requires closed headphones that are equipped with microphones that lie in the auditory canals.

Shown in Figure 6 is a schematic block diagram of an audio system SYS or an A/V system. In the present case, an acoustic reproduction via 6 channels (Dolby 5.1) with 6 loudspeakers SPE or via stereo headphones HPH (2 channels) is sketched in. In principle, the invention can be used for any number of channels.

The audio system SYS shown here comprises a series of components that do not all need to be present in this combination, and in certain cases can be present only individually.

A set-control unit SET with corresponding memory RAM, ROM takes over the fundamental control tasks and is connected to an input unit INP of the audio system SYS, as well as to an infra-red receiver IR for communication with a remote control. The set- control unit SET also activates a display DIS of the audio system SYS.

The set-control unit SET also controls the algorithm to determine the user- specific auditory thresholds Th(f) per channel, and the determination of the original sound level setting VoI(OL) as described above.

Also shown are a real-time data source RTD, such as for example a tuner, internet etc., as well as a further data source DST, for example a playback unit for CDs, DVDs, a hard disk, minidisk, etc. The set-control unit SET likewise activates these two sources RTD and DST.

These two sources RTD, DST are, like an external microphone EMI that is also shown, (this latter via a microphone amplifier MAM), connected to a unit SSE for selecting the data source.

This unit SSE is in turn connected, either directly for digital data or for analog signals via an A/D converter KON, to a digital signal processing arrangement DSP, here in the form of a digital signal processor, this signal processor supplying audio signals via an amplifier AMP to the six loudspeakers SPE or via a further amplifier AHP to the headphones HPH.

The algorithm, described above, for the amplification of audio signals Sig(f) according to the invention is executed here by the digital signal processing arrangement DSP, this signal processing arrangement DSP being programmed accordingly for this.

The necessary data in respect of the auditory threshold Th(f), which are determined by means of the set-control unit SET, are advantageously stored in the memory RAM of the set-control unit SET. Alternatively, they can of course be stored in the memory RAM of the signal processing arrangement DSP, or another accessible memory in the audio system SYS. In the case of audio systems SYS that can be connected to the internet, for example in order to thus download audio files onto the audio system SYS, it can be provided that the corresponding data relating to the auditory threshold Th(f) are loaded into the audio system SYS from a memory that is accessible via the internet. If in the case of the audio system SYS that is shown in Figure 6, one assumes that it is a "conventional" familiar audio system, then for the realization of the invention, in principle "only" software changes are necessary for the digital signal processing arrangement DSP for the amplification of the audio signals as well as for the production of test sequences for determining the auditory threshold, as well as changes to the set-control unit SET, in particular to control the test procedure and in relation to the user interface. The means of the signal processing arrangement DSP for executing the method described above, for amplifying audio signals, are thus specific software or an expansion of existing software.

In principle, the invention can be used in the case of any system for the reproduction of audio signals, including in the case of A/V systems, as already mentioned. It is favorable if the signal processing arrangement DSP is a constituent part of the audio system SYS.

In principle, it can be provided that the signal processing arrangement for implementing the algorithm according to the invention is designed separate from the audio system, as a separate or external unit, this external unit then being connected to the audio system for the purposes of the invention. Expediently, this external unit is then additionally set up for determining the individual auditory threshold of a user.

In this latter case, various possibilities are conceivable, which are briefly explained on the basis of Figure 6: a) The external unit for the method according to the invention comprises a system SYS as shown, in which however the sources RTD, DST are omitted and are replaced by an analog and/or digital input, to which a conventional audio system or A/V system can be connected. b) The variant described under point a) can be further reduced, in that the amplifier unit AMP, AHP are also omitted. In this case, the output of the digital signal processing unit DSP is connected to the input for example of an existing audio amplifier.

What is significant here is that the volume control is affected via the external unit and not via the existing amplifier. c) The variant described in point b) can be optimized for mobile headphone-operated devices, in that the units DST, RTD and SSE are left out. Instead, the external unit is connected to the headphone output of a mobile audio device (or preferably to the digital output, if present). In the case of this embodiment, the amplifier unit AMP does not apply. However, downstream of the digital signal processing arrangement DSP is a headphone amplifier AHP, so that the headphones HPH can be connected directly with the external unit.

Claims

1. A method for the frequency-dependent amplifying of the sound level of an audio signal, in which method the audio signal (Sig(f)) is divided, at least in a certain frequency range, into a number (n) of frequency bands (fl ...fl 0), and a particular auditory threshold (Th(fl)...Th(flO)) is specified for each frequency band

(fl...flθ), and for each frequency band (fl...flθ) the sound level of the audio signal (Sig(f)) in the frequency band (fl...flθ) is compared with the respective pre-determined auditory threshold (Th(fl)...Th(flO)), and the audio signal (Sig(f2), Sig(f3), Sig(f4)) in those frequency bands (f2, β, f4) in which the sound level lies just below the respective auditory threshold (Th(O), Th(f3), Th(f4)) is amplified in such a way that the sound level of the amplified signal (Sig(f) + Amp(f)) in these frequency bands (f2, β, f4) lies above the auditory threshold (Th(f)).

2. A method as claimed in claim 1, wherein furthermore the audio signal (Sig(f2), Sig(β), Sig(f4)) in those frequency bands (f2, β, f4) in which the sound level lies at or just above the respective auditory threshold (Th(O), Th(β), Th(f4)) is amplified.

3. A method as claimed in claim 1 or 2, wherein the audio signal (Sig(O),

Sig(β), Sig(f4)) in different frequency bands (O, β, f4) is amplified more or less strongly depending on the sound level of the audio signal (Sig(f)) in relation to the auditory threshold (Th(f)).

4. A method as claimed in any one of the claims 1 to 3, wherein for each channel of the audio signal (Sig(f)), a corresponding auditory threshold (Th(f)) is specified and the amplification of the audio signal is carried out separately for each channel (Sig(f)).

5. A method as claimed in any one of the claims 1 to 4, wherein the audio signal (Sig(f)) is amplified in all frequency bands in which the sound level lies below a resulting threshold value (Th(f) + ML).

6. A method as claimed in claim 5, wherein the audio signal (Sig(f)) is amplified most strongly in those frequency bands in which the sound level lies most clearly below a resulting threshold value (Th(f) + ML), and the amplification decreases with a decreasing difference between this resulting threshold value (Th(f) + ML) and the sound level of the audio signal (Sig(f)) in a frequency band.

7. A method as claimed in claim 6, wherein the amplification of the audio signal (Sig(f)) decreases as the auditory threshold (Th(f)) in a frequency band is increasingly exceeded, and becomes zero when the resulting threshold value (Th(f) + ML) is reached and exceeded.

8. A method as claimed in any one of the claims 1 to 7, wherein after selective amplification of the audio signal (Sig(f)) in certain frequency ranges (f2, f3, f4) has taken place, an even reduction of the amplified signal (Sig(f) + Amp(f)) takes place in all frequency ranges (fl...flθ).

9. A method as claimed in claim 8, wherein the reduction of the amplified signal (Sig(f) + Amp(f)) is selected in such a way that the overall spectral energy (SE) of the audio signal (Sig(f)) that is perceived by a user essentially assumes the same value before and after amplification.

10. A method as claimed in any one of the claims 1 to 9, wherein the width of the frequency ranges (fl...flθ) is selected such that it lies in the order of magnitude of the frequency resolution capability of a human ear.

11. A method as claimed in any one of the claims 1 to 10, wherein the maximum amplification (MA) of the audio signal (Sig(f)) becomes zero at and above the original sound level VoI(OL) of the audio signal (Sig(f)), and at a lower sound level of the audio signal (Sig(f)) the maximum amplification (MA) rises up to a pre-determined maximum value (MAmax) for the maximum amplification (MA).

12. A method as claimed in any one of the claims 1 to 11, wherein the pre- determined auditory threshold (Th(f)) for different frequency ranges (fl...flθ) is established by determining the individual auditory threshold of a person.

13. A method as claimed in claim 12, wherein the individual auditory threshold (Th(f)) is determined separately for each channel (Sig(f)) of the audio signal.

14. A method as claimed in claim 12 or 13, wherein when determining the individual auditory threshold (Th(f)) as a function of the frequency (f), otoacoustic emissions are also taken into account.

15. A signal processing arrangement (DSP) with means for implementing a method as claimed in any one of the claims 1 to 14.

16. An audio system (SYS) with a signal processing arrangement (DSP) as claimed in claim 15.

17. An audio system (SYS) as claimed in claim 16, which comprises means (SET, SPE; HPH) for determining the auditory threshold (Th(f)) of a user.