CA2154883A1 - Digital audio limiter - Google Patents
Digital audio limiterInfo
- Publication number
- CA2154883A1 CA2154883A1 CA002154883A CA2154883A CA2154883A1 CA 2154883 A1 CA2154883 A1 CA 2154883A1 CA 002154883 A CA002154883 A CA 002154883A CA 2154883 A CA2154883 A CA 2154883A CA 2154883 A1 CA2154883 A1 CA 2154883A1
- Authority
- CA
- Canada
- Prior art keywords
- output
- peak
- signal
- gain
- amplitude
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Abandoned
Links
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03G—CONTROL OF AMPLIFICATION
- H03G7/00—Volume compression or expansion in amplifiers
- H03G7/007—Volume compression or expansion in amplifiers of digital or coded signals
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04H—BROADCAST COMMUNICATION
- H04H60/00—Arrangements for broadcast applications with a direct linking to broadcast information or broadcast space-time; Broadcast-related systems
- H04H60/68—Systems specially adapted for using specific information, e.g. geographical or meteorological information
- H04H60/71—Systems specially adapted for using specific information, e.g. geographical or meteorological information using meteorological information
Abstract
The invention relates to limiting the peak amplitude of an audio signal in one or more frequency subbands while preserving the apparent loudness. Applications such as a Studio-Transmitter Link (STL) for broadcasting sometimes use perceptual coding to deliver an audio signal originating from a studio to a broadcast transmitter. The peak amplitude of the audio signal will have been limited by means of a limiter or otherwise, and a perceptual encoder reduces the informational capacity requirements of the audio signal for transmission across a link to a broadcast transmitter. A perceptual decoder receives the coded signal from the link and reproduces the audio signal for the broadcast transmitter. The peak-amplitude of the signal reproduced by the perceptual decoder may sometimes exceed the capabilities of the broadcast transmitter even though the peak amplitude of the audio signal input to the perceptual encoder is properly limited. This increase in peak level is referred to as "peak-level increase" or PLI. Transmitter overload resulting from PLI can create audible distortion and/or impermissible broadcast conditions such as excessive FM deviation. Various embodiments of apparatus and method are described which estimate PLI caused by signal processing such as perceptual coding and which apply corrective gain to the portions of the audio signal bandwidth so as to limit peak amplitude while preserving apparent loudness.
Description
~154883 DESCRIPTION
DIGITAL AUDIO LIMITER
Technical Field The invention relates in general to limiting the peak amplitude of an audio signal.
5 More particularly, the invention relates to limiting the peak amplitude of one or more frequency subbands of an audio signal while preserving the apl)afe-lt loudness.
B~l k~round Art There is considerable interest among those in the fields of bro~-lc~cting and recording to reduce the amount of information required to transmit or record an audio signal intendeA
10 for human perccpLion without degrading its subjective quality. Analog signals with reduced informational require",ent~ can be carried within narrower bandwidths, and digital signals with reduced information requilc,,,cnts can be carried at lower bit-rates.
One co"""on technique used to reduce the info""ational ~cquil~ ents reduces the dynamic range of the audio signal to be transmitted or recorded. Dynamic range control is 15 used to protect euuip,,,. nl from excessively high-amplitude signals and to achieve certain artistic results. In general, the overall goal of dynamic range control is to alter the dynamic range of an audio signal without introducing any other ~rcepLible distortion.
In bro~ç~cting, for example, the dynamic range of audio signals is controlled prior to broadcast to avoid overloading tr~ncmiccion equipment and/or to avoid severe audible 20 distortion. Similar concel"s apply to conventional tape and disc rccofdillgs. The control of dynamic range may be accomplished by automatic techniques such as "CO"~pf ssion" and "limiting" or by appropfiate manual gain setting~c.
Dynamic range "col~ples~ion" reduces the dynamic range of an input signal by applying a variable gain factor varied in response to input signal amplitude. A "co",p~ssor"
25 provides for signal amplitude dynamic range co"~prcssion A "limiter~ provides "limiting"
which is a special case of co"~plession, preventing the peak-amplitude of an input signal from -Yce~Aing a specified level by applying a very low gain factor to high-level signals.
A second common technique used to reduce the informational l~uire~ en~s of an audio signal reduces the amount of information used to represent or code the audio signal; however, 30 as the amount of information is reduced, encoding inaccuracies increase and may become audible in the form of "coding noise." Coding noise degrades the subjective quality of the coded signal. So-called psychoacoustic-coding or pclccplllal-coding techniques attempt to reduce the informational requirements of an audio signal without introducing audible coding noise. Two examples of "split-band" pclceplllal-coding techniques are subband coding and transform coding. Perceptual-coding techniques exploit a characteristic of human hearing; a stronger signal may mask or render inaudible a weaker signal if the two signals are sufficiently 5 close in frequency. By splitting an audio signal into narrow frequency bands and independently coding the signal energy in each band, the aural effect of the coding noise is more likely to be inaudible because it is confined to the same frequency band as the coded spectral energy.
Coding systems which implement ~rce~ut~lal-coding techniques attempt to reproduce 10 a ~p~ ;on of the input audio signal which preserves the perceived loudness of input signal spectral co~pon~ . This is often accomplished by preserving some measure of spectral amplitude such as root-mean-square (RMS); however, many ~r~plual-coding systems lead to uncertainties in the peak-amplitude level of the reproduced signal. These uncertainties may include increases in peak-amplitude, I~;îel.ed to herein as "peak-level increase" or PLI, 15 which is tolerable in many coding applications where it is in~u~ible The Studio-Tr~ncmitter Link (STL) for bro~lc-~cting~ which delivers an audio signal origin~ting from a studio to a broadcast tr~ncmitter~ is one example of an application which cannot tolerate excessive amounts of PLI. In one embodiment, an STL includes a colnplel,sor and limiter which reduce the dynamic range and limit the peak-amplitude level of an audio 20 signal, a ~r~eptual-encoder which reduces the informational capacity requirements of the audio signal, a communications ch~nntol for delivering the encoded signal, and a p lcepl~lal-decoder for reprodllçing the co,np~ss~ and limited audio signal for subsequent bro~ ct~
ReC~u-ce of PLI, the peak-amplitude of the signal reproduced by the pe~eptllal-decoder may sometimeS exceed the capabilities of the broadcast tr~ncmittPr even though the peak amplitude 25 of the audio signal input to the pel~e~ual-encoder is pfoyelly limited. Tr~nsmitter overload resulting from PLI can create audible distortion and/or impermissible broadcast conditions such as excessive FM deviation.
Known techniques for controlling PLI include clippels, inct~nt~n~us gain reduction amplifiers, and conventional wide-band limiters. Unfortunately, these techniques introduce 30 undesirable audible distortion in the reproduced signal. Clippers generate excessive harmonic distortion. Inct~nt~ne~us gain reduction amplifiers, in effect, smear spectral colllponents in the frequency domain. Conventional wide-band limiters reduce the perceived loudness of the reploduced signal.
wo 94/19883 215 ~ ~ 8 3 PCT/US94/01639 Disclosure of Invention It is an object of the present invention to provide for a signal ploce~ing system in which both al)pa ent loudness of the reproduced audio signal and PLI are controlled within acceptable levels.
It is a further object of the present invention to provide for a signal processing system in which a high-quality leprese,-td~-on of the input audio signal is reproduced. These objects are achieved by the invention as claimed.
In accorddnce with the te~^hings of various aspects of the present invention in one embodiment, a signal pr~ceccing system receives a peak-amplitude limited input audio signal, generates a plocessed audio signal in response to the input audio signal such that peak-level increase may be present, estim~tes the peak-level increase of the full-bandwidth processed audio signal, and generates an output audio signal by applying to the portion of the full-bandwidth subject to PLI a gain factor adapted in response to the estim~tçd peak amplitude.
In accoldance with the tç~ ingc of another aspect of the present invention, a signal proceccing system receives a peak-amplitude limited input audio signal, generates a processed audio signal in response to the input audio signal such that peak-level increase may be present, estim~tçs the peak-level increase of the full-bandwidth processed audio signal, and gen~,dtes an output audio signal by applying to portions of the full-bandwidth subject to PLI a plurality of gain factors adapted in response to the estinl~tçd peak amplitude.
The present invention is generally applicable to signal proceccing which is subject to PLI. As ~iiscussed above, many p~,~;t;plual-coding systems and methods are known in the art which gel-cldte, in ~Jonse to a peak-amplitude limited input audio signal, a plOCf ~ed audio signal with PLI. Of cignifi~nce to practical embodiments of the present invention, the inventors have determined empirically that in broadcast STL applications, the predo."inant audible effects caused by PLI generated by pe~cepl~lal-coders are confined to high-frequencies above about 5 kHz. It is therefore possible to adequately correct for PLI by applying an appl~pliale gain factor to only those high-frequency signal components.
A low-pass filter (LPF) with a frequency-dependent phase shift is another example of signal processing which can preserve app~G--t loudness but permit PLI. For example, such 30 a 20 kHz LPF receiving a 6 kHz square wave input signal will pass the third harmonic at 18 kHz with its phase shifted relative to the phase of the fund~mçnt~l frequency. If the relative phase shift is of an a~o~liate amount, the third harmonic will combine with the fun-l~m~o-n~l frequency to increase the peak amplitude rather than to flatten the waveform and reduce the 215~88~ 4 peak amplitude. Although the a~arent loudness of the low-pass filtered signal may be preserved, the peak-~mplitude of the passed signal will increase.
The present invention is also applicable to practical implementations of co",ponents such as digital-to-analog converters and anti-aliasing filters which generate PLI. Other applications will be known to those skilled in the art.
Various aspects of the present invention and its plefelled embodiments are set forth in greater detail in the following "Modes for Carrying Out the Invention" and in the acco",panying drawings. It should be appreciated that the following liccll~cion sets forth several embo lim~nt~ by way of example only, and that these examples are not intended to ~resent any limitations in application or implementation.
Although the following ~is~uccion is more particularly dir~c~ed toward digital split-band coding within an STL application, the present invention is not so limited. As discu~ briefly above, aspects of the present invention are applicable to a broader variety of signal proce~sing systems. Furth~.",~"t;, the present invention may be embodied in systems implemented by analog techniques as well as digital techniques.
Brief D~ t.on of Drawin~c Figure 1 is a functional block diagram illustrating the structure of one embo iimçnt of an audio signal limiter according to various aspects of the present invention.
Figure 2 is a functional block diagld", of an STL incorporating an embodiment of the present invention.
Figure 3 is a functional block diagram of one embodiment of a control system according to various aspects of the present invention.
Figure 4 is a functional block diagram of one embodiment of a peak-amplitude estim~tor.
Figure S is a hypothetical graphical re~ sen~ation of one peak hold function.
Figure 6 is a functional block diagram of one embodiment of a control system according to various aspects of the present invention.
Figure 7 is a functional block diagram illustrating an alternative band-splitting structure within an embodiment of an audio signal limiter according to various aspects of the present invention.
Figure 8 is a functional block diagram illustrating the structure of one embodiment of an audio signal limiter according to various aspects of the present invention.
wo 94/19883 215 4 8 8 3 PCT/US94/01639 _ 5 Figure 9 is a functional block diagram illustrating an alternative hybrid input-controlled/output-controlled structure according to various aspects of the present invention.
Figures 10a-lOc illustrate functional block diagrams of three types of limiters.
Modes for Carryin~ Out the Invention S A. Intro~-c~ic~l Figures 10a-lOc illustrate functional block diagrams of three types of limiters. The limiter shown in Figure 10a is an output-controlled limiter which uses negative fee~ibacl~.
Gain element 1004 receives an input signal from path 1002 and generates an output signal along path 1006 by applying a variable gain factor to the input signal. Control system 1008 receives the output signal from path 1006 and passes a gain control signal along path 1010 which selects the gain factor used by gain element 1004. Output-controlled limiters are very tolerant of variations in the ch~dclelistics of the control system and the gain elem~-nt~ but they cannot avoid ovelsl-oot in the output signal because of the inherent delay in the control system.
The limiter shown in Figure 10b is an input-controlled limiter. Delay element 1024 receives an input signal from path 1022 and passes the delayed signal along path 1026 to gain element 1028. Control system 1032 receives the input signal from path 1022 and geneldles a gain control signal along path 1034 which selects the gain factor used by gain elemçnt 1028.
Gain element 1028 receives the delayed input signal from path 1026 and gene.~dtes an output signal along path 1030 by applying a variable gain factor to the delayed input signal. Input-controlled limiters require a control system and a gain element with precisely known char~ct~ri~tics. This is usually much easier to implement with digital technologies than it is to implemçnt with analog technologies; however, input-controlled limiters can avoid overshoot in the output signal by using a delay elen ent The limiter shown in Figure 10c is a hybrid of the other two types of limiters. Delay elçmçnt 1044 receives an input signal from path 1042 and passes the delayed signal along path 1046 to gain element 1048. Gain element 1052 receives the input signal from path 1042 and geneldtes a quasi-output signal along path 1054 by applying a variable gain factor to the input signal. Control system 1056 receives the ~uasi-output signal from path 1054 and gene,~es a gain control signal along path 1058 which selects the gain factor used by gain element 1052 and gain element 1048. Gain element 1048 receives the delayed input signal from path 1046 and g~ t~s an output signal along path 1050 by applying a variable gain factor to the delayed input signal. The hybrid limiter does not require a control system and gain elements with precisely known characteristics, but it does require two gain elements with precisely WO 94/1s883 PCTIUS94/01639 matched characteristics. This is usually much easier to implement, even with analog technologies. Furthermore, by using the delay element, the hybrid limiter is also able to avoid overshoot in the output signal.
Figure 2 is a functional block diagram of a portion of a broadcast system including an S STL incoll~o,~ling one embodiment of the present invention. The following discussion more particularly describes various embodiments of the present invention for this application. The embodiments described below incorporate an input-controlled~limiter, but it will be readily a~)pa,~ nt in light of the prior art that these different embQdiments may be mo lified to also incorporate an output-controlled limiter or a hybrid limiter as described above.Referring to Figure 2, co~ )ressor/limiter 204 receives a studio signal from path 202 and generates a peak-amplitude limited re~ sent~tion of the studio signal which it passes to encoder 206. Encoder 206 generates an encoded signal which it passes along communication path 208. Decoder 210 receives the encode~ signal from communication path 208, reproduces the colllp.essed and limited signal, and passes the reproduced signal to limiter 212. Limiter 212 co~,~cls any PLI in the reproduced signal and passes the col,ecled signal along path 214 to a broadcast tr~ncmitter. It should be understood that co,llpl~ssor/limiter 204 does not form part of the present invention and is not required to practice the present invention. For example, app,~iate manual gain settings alone may be sufficient to bound signal dynamic range; however, collll)~ssion is sometimes used to reduce dynamic range and limiting is sometimes used to prevent ac~idPnt~l violations of prescribed peak amplitude limits.
For ease of ~1iscu~ion~ pr~re.led digital embodiments of the present invention for an FM-broadcast STL application are desclibed which assume that the reproduced audio signal delivered to limiter 212 has a 15 kHz bandwidth and is sampled at a rate of 44.1 kilo-samples/sec. Changes required to implement an embodiment of the present invention for use with signals having different characteristics are within the abilities of those having ordinaly skill in the art.
B. Basic Structure Figure 1 illustrates the structure of one embodiment of an input-controlled limiter according to various aspects of the present invention. Delay element 104 receives an input signal from path 102 and passes the delayed signal along path 106 to splitter 108. Splitter 108 splits the delayed signal into at least two sub-signals, passing a first sub-signal along path 112 to gain element 114 and passing the rem~ining one or more sub-signals along path 110 to colllbiner 118. Control system 122 receives the input signal from path 102 and generates a W O 94/19883 215 4 8 8 3 PCTrUS94/01639 _ - 7 -gain control signal along path 124 which selects the gain factor used by gain element 114.
Gain element 114 receives the first sub-signal from path 112 and generates a second sub-signal along path 116 by applying a variable gain factor to the first sub-signal. Combiner 118 generates an output signal along path 120 by combining the sub-signals received from paths 1 10 and 1 16.
Splitter 108 divides the delayed input signal into two or more sub-signals leplcsenting two frequency subband col"ponents. The first frequency subband component is subject to PLI. The second frequency subband colnponent includes that portion of the s~ , Lllll in which PLI does not exist. The first frequency subband component which may have PLI is ~ple~nted in Figure 1 by the first sub-signal passed along path 112 to gain element 114. The other frequency subband component is lcpresented in Figure 1 by the one or more sub-signals passed along path 110 to colllbiner 118.
1. Signal Pr~ceccing Path The structure shown in Figure 1 has two paths. A signal procesc;ng path includesdelay 104, splitter 108, gain element 114, and combiner 118. A se~,a.dte control path, below, includes control system 122.
Delay elernent 104 may be implemented by any technique applupliate for delaying the audio signal received from path 102. The duration of the delay is usually set subsl~n~ y equal to the length of time required for control system 122 to respond to PLI requiring 20 correction. One method which may be used to establish this duration is to establish the delay which yields the maximum cross-correlation score between the delayed signal on path 112 cont~ining PLI and one minus the gain factor (1-g) generated by control system 122.
Splitter 108 may be implemented by any technique for dividing a signal into frequency subbands including, but not limited to, analog filters, digital filters, and so-called frequency-25 domain transforms.
Combiner 118 may be implemented by any technique for combining the sub-signals into a full-bandwidth signal. The combiner implementation technique will normally be the inverse of the implement~tion technique chosen for splitter 108.
Gain element 114 may be implemented by any technique a~r~liate for the signal 30 received from path 112. For Py~mple, operational amplifiers may be used in analog systems and m-lltipli~tion or scaling may be used in digital systems. The resolution and range of the gain element should be s~lected in conjunction with the operational characteristics of control system 122, discussed below.
WO 94/19883 215 ~ 8 8 3 PCT/US94/01639 2. Control Path One embodiment of control system 122 is shown in Figure 3. Referring to Figure 3, splitter 303 splits the input signal received from path 302 into two frequency subband co",~nents in the same manner as that performed by splitter 108 discussed above. The second frequency subband component, which is a~cumed not to contain PLI, is passed to peak estim~tor 305. Peak estim~tor 305 estim~t-Ps the peak ampl~itude of the second frequency subband component and passes the estim~te to peak hol~ 307. Peak hold 307 passes to threshold co",l,~rc 308 a pulse-shaped signal which holds the estim~t~d peak amplitude for a specific period of time.
The first frequency subband c(jmponent, which may contain PLI, is passed to peakestim~tor 304. Peak Pstim~tor 304 estimates the peak amplitude of the first frequency subband co"~ponent and passes the ectim~t~p to peak hold 306. Peak hold 306 passes to threshold CG~pale 308 a pulse-shaped signal which holds the estim~ted peak amplitude for a specific period of time.
Threshold col"palc 308 obtains a peak-amplitude estim~te of the full-bandwidth input signal by col"bining the two peak hold signals and cG,npales the e~ te with a reference level. Threshold co"")are 308 passes the two peak hold signals and the results of the cGI"pdlison to gain select 310. Gain select 310 selects a gain factor in les~n~ to the signals received from threshold CGIllparC 308 and passes the sPle~ted gain factor to control filter 312.
Control filter 312 generates along path 314 a gain control signal in respon~ to the select~P~
gain factor. The functions rcp-~3ented in Figure 3 affect both the steady-state and the dynamic chalac~llstics of the control system.
The following rliscn~ion of the embodiment shown in Figure 3 is directed toward digital implementations; however, it should be applcciated that COIl~ s~onding functions may be implP-m~nted using analog techniques.
a. Peak F.ctim~tor In a plcrelled embodiment of the present invention for an FM-broadcast STL, the first frequency subband col~pol-ent contains high-frequency spectral co,n~nent~ up to about 15 kHz. Peak esli...~tor 304 uses an "upsampling" filter to interpolate the samples of the first 30 frequency subband component, thereby improving the accuracy of the peak-amplitude estim~t~p.
~lthough samples taken at or above the Nyquist rate permit accurate recovery of a sampled signal, it is well known that the samples themselves do not accurately r~le3ent the signal peak amplitude. In many digital applications the only conceln is the ability of the digital system to process and transmit the samples themselves. In applications such as a broadcast STL, wo 94/19883 215 4 8 8 3 PCT/US94/01639 _ g however, the actual peak amplitude of the analog signal reproduced from the digital signal is of concern. There~ore, the peak estim~tor function interpolates between samples to obtain a more accurate çstim~te.
In the same p~cfelred embodiment, the second frequency subband component contains 5 only frequencies below about 5 kHz. An accurate peak amplitude can be estimated accurately from samples taken at a rate of 44.1 kHz; therefore, peak estim~tor 305 does not need to interpolate the samples of the second frequency subband component.
Figure 4 shows a functional block diagram of a preferred embodiment of an interpolating peak estimator. Upsampler 404 and upsampler 406 are each two-times10 upsampling filters, and together the two upsamplers generate four samples in response to each input signal sample received from path 402. Each u~ --pler can be implem~nt~l efficiently as a low-pass half-band FIR filter. More particularly, for the FM-broadcast STL application, filter characteristics are chosen so as to obtain a 15 kHz p~sb~nd with -40 dB of attenuation in the stopband The filter for upsampler 404 comprises seventeen taps and upsamples to a rate of 88.2 kilo-samples/sec. The filter for upsampler 406 comprises nine taps and upsamples to a rate of 176.4 kilo-samples/sec.
Selector 408 co,n~ares the four most recent interpolated samples and selects the one with the largest absolute value. Selector 410 coml)~es the magnitude selected by selector 408 with a stored value. Selector 410 passes the larger of the two values along path 412 and saves the value s~lected by selector 408 as the new stored value.
Conceptually, the peak estim~tor shown in Figure 4 up~rnples the input signal by four times and, for each set of four interpolated samples, selects the largest absolute value from the eight most recent interpolated samples.
b. Peak Hold The functions r~plesented by peak hold 306 and peak hold 307 shown in Figure 3 are nPcçsc~ry because the limiting action of control system 122 and gain element 114 shown in Figure 1 is not instantaneous. In order to minimi7~ the audible effects of the limiting action within the signal procescing path, it is illl~ t to control the rate at which the gain factor applied by gain element 114 is changed. Various considerations are discussed below. It is 30 sufficient at this point to realize that the gain factor should not be changed in~t~nt~neously.
Therefore, unless some col-,pensation is made, the control system will not be able to adequately respond to short-duration PLI intervals. The function r~l senled by peak hold 306 and peak hold 307 in Figure 3 is one way in which compensation may be made.
Wo 94/19883 215 ~ 8 8 3 PCT/US94/OlG39 Each peak hold function generates a pulse signal of duration H having an amplitude equal to the peak-amplitude estim~t~- received from the peak estim~tors. For example, in response to a peak-amplitude estim~te of xl at time t" the peak hold function generates a pulse signal with an amplitude equal to xl, starting at time t, and ending at time tl +H. If a larger S peak-amplitude estim~te x2 is received at time t2 before the first pulse ends, however, the peak hold function imme~i~tely generates another pulse signal with an amplitude equal to x2 of duration H starting at time t2 and ending at time t2 +H. If a ~eak-amplitude estim~te X3 smaller than x2 is received during the current pulse signal at time~t3, the peak hold function generates a pulse signal with amplitude equal to X3 starting at the end of the current pulse signal and 10 ending at time t3+H. Figure 5 provides a hypothetical graphical ~ esentation of this function with a hold duration of 4T.
The following pseudo-code program seg.,l.~nt illustrates the logic of an embodiment which implements the peak hold function described above and illustrated in Figure 5.
i--O
Y--O
X(O:H)--O
while TRUE
X(i)--x(n) if Y ~ X(i+1) then Y--ms~(Y,X(i)) else Y--max(X(I~,,X(i+H-l)) endif y(n)--Y
i--i+l endwhile where x(n) = peak-~mplitude estim~t~- at time n, H = the peak hold duration, and y(n) = peak hold function output at time n.
The program segment implements a circular buffer X of length H; therefore it should be understood that the variable i is an index or pointer to the buffer and the notation i+k r~,csenls the eA~,- ssion (i+k) - int( i k~ H
where int(q) = the integer portion of q.
After initi~li7ing the variables i, Y and circular buffer X, the program se.gment contained within the "while" group is r~peLili~rely executed. As each peak-amplitude estim~te wo 94/1g883 215 4 8 8 3 PCT/US94/01639 x(n) at time n is received, its magnitude is stored in circular buffer X. The current value of Yis co,-,palcd to the value stored H time periods earlier in buffer X. If the two are not equal, Y is set equal to the larger of the magnitude just stored in buffer X or the current value of Y.
If they are equal, this indicates that the peak value has been held for duration H and Y is set equal to the largest of the last H magnitudes stored in buffer X. Many other implementations are possible.
The length of duration H is a design choice which should be made together with other design choices affecting the gain select and the control filter functions, discussed in more detail below.
c. Threshold Compare Threshold co-~palc 308 co,npa,es an çstim~te of the full-bandwidth input signal peak amplitude with a reference level and passes the results of the comparison to gain select 310.
In an embodiment according to Figure 3, an estim~tç of the full-bandwidth input signal may be obtained by combining the peak hold signals from peak hold 306 and peak hold 307.
In a p~cr~llcd embodiment of the present invention for broadcast STL applications such as that shown in Figure 2, the reference level is pre~efined according to the peak-amplitude limit i~.pose~ by cG",pl(ssor/limiter 204 upon the studio signal. The threshold co,np~e reference level is the desired peak-amplitude limit of the signal reproduced by pe,eept~al-decoder 210.
In principle, the r~f~ence level need not be predefined but could be provided by an external signal. The desired peak-amplitude level could be passed with the signal having PLI.
Referring to Figure 2, for example, the desired peak-amplitude limit could be encoded by ~ncodçr 206, passed with the encoded signal along path 208, extracted by de~oclçr 210 and provided to limiter 212. Many variations are possible.
d. Gain Select Gain select 310 establishes a gain factor in response to the output of threshold co"~pa,e 308. If the peak-amplitude estim~te of the full-bandwidth signal is less than the reference level, a gain factor equal to one is selected. If the peak-amplitude estim~tç ~xcee~ the reference level, however, a gain factor is selected which effectively reduces the amplitude of the frequency subband co",ponent con~ining PLI such that the peak amplitude of the output signal is reduced to the desired amplitude defined by the reference level. Additional details ~,~ining to the design of limiters may be found in Bosi, "Low-Cost/High-Quality Digital Dynamic Range Processor," AES 91st Convention, New York, October 1991, Preprint 3133, which is incol~o,~ted by reference in its entirety.
WO 94/lg883 21 S 4 8 8 3 PCT/US94/01639 For conventional full-bandwidth limiters~ a gain factor g may be established at time n subst~nti~lly in accordance with [7h - 20 bg ~(n)~
where g(n) = selected gain factor at time n, S Th = reference level e~p,lcssed in dB, x(n) = peak-amplitude estim~tt~ at time n, and R = cG,~plession ratio. .; ~ `
Alternatively, a gain factor g(n) may be obtained from the ratio between the reference level Th and the peak-amplitude estim~te x(n). This is equivalent to the results obtained from 10 equation (1) where the co~ ,rcs~,ion ratio R is infinitely large.
By using equation (1) and setting the co"~p,ession ratio R at some finite value, gain select 310 is able to select a gain factor which balances PLI reduction with the decrease in appar~l loullnrss of the output signal. For many broadcast STL applications, a cc""~,lession ratio R=100 effectively removes PLI.
For ~,u~ oses of the present invention, however, the gain factor c~lcul~ted acco.~ing to equation (1) may not provide nearly enough limitjng if equation (1) is applied to only a single peak-amplitude estim~t~. At least two peak-amplitude estim~t~s must be used in applications such as the broadcast STL iiccllc~d above where PLI is confined to certain frequency bands which do not rep,esent the p~edol..ill~nt spectral energy for the full-bandwidth signal. Neve. Iheless, it is useful to first describe a simpler gain select function which ~ccllmes that the frequency band cont~ining PLI l~iesents subst~nti~lly the full-bandwidth signal energy. For ease of ~liccllcsion~ this gain select function is ~felred to herein as "simplex gain select" because it requires only one input source, the çstim~ted peak amplitude of the frequency subband cG-~-ponent containing PLI.
i) simplex gain select Although equation (1) defines the desired gain to select, a straight-forward calculation is computationally intensive. In a pç~felled embodiment, a more effici~nt process is implçmçnt~ by selecting the desired gain from a pre-computed gain table. The gain table T
contains a set of gain factors which may be defined by the following pseudo-code program sçgm~nt WO 94/19883 215 ~ 8 8 3 PCT/US94/01639 __ - 13 Th---2 R--lOO
A",p--16 S S--(R-l) I R
Am~ ~ 2 -l A ~ lo~/20 A
Tsz _ (A",~" - A,b) / Ad,p ~--O
A--A,b for i from O to Tsz l T~l--10~-~/20 A . A + A"~p ~--Th -201cgA
endfor where N = number of bits available for digital lcplcscntation, Th = reference level in dB, R = CGIllp~ Sion ratio, A~ep = amplitude step siæ, A"",~ = maximum digital le~fe~n~;1tion, A,h = absolute reference level, Tsz = gain table size, and T[il = gain factor i in the gain table.
In one embo~iment sixteen bits are used to express signal amplitudes, Ihe1cforc, N is set equal to l6. It should be appleciatcd that the gain table T contains anti-log gain factors, the~cfo~c the gain increment a between ~ cent table entries is not constant. The amplitude step size A~ep is selected such that the largest gain increment ~\ between ~ Cçnt table entries is not too large. In a p1cfe11~ emb~im~-nt, A~ep is set to sixteen which ensures that /~ is always less than 0.05 dB.
In the ylcfe11~ embodiment, gain select 3lO receives from threshold co",parc 308 a signal in(~ ting the difference D between a peak-amplitude estim~te y(n) and the 1efelcnce level A,h. Gain select 3 lO uses D as an index into the gain table T and thereby selects the gain factor g al,pro~liate to reduce the PLI causing the pcak amplitude error, or g(n) = Tly(n)-A~h~. (2) Many other implementations of the gain select function are possible.
ii) duplex gain select In applications having PLI oc-;u11ing in frequency subband co",l,onc1-ts which have significantly less than the full-bandwidth spcctral energy, simplex gain select does not provide enough limiting to optimally reduce PLI to desired levels. A more optimum gain select function, r~fe"~ to herein as "duplex gain select," requires two peak-amplitude estim~tps.
The conceptual basis for duplex gain select assumes that the sum of the peak amplitudes in various frequency subbands equals the peak amplitude of the full-bandwidth input signal.
For example, two bandwidth limited signals having peak-to-peak amplitudes of 0.4 volts and 0.6 volts, respectively, when combined, will form a signal having a peak-to-peak amplitude of one volt. Although this assumption is not precisely correct, it is accurate enough for the purposes served by the present invention in many àpplications. lmproved accuracy can be achieved by using an alternative embodiment to that illustrated in Figure 3. For example, peak estim~tes received from peak estim~tors 304 and 305 could be combined, passed to one peak hold c~",ponent and subsequently processed by threshold compare 308.
In an embodiment of the present invention for applications such as the broadcast STL, the signal received from peak hold 307 re~,resents the peak amplitude of frequencies below about S kHz. The peak amplitude estim~te for frequencies above about S kHz is received from peak hold 306 via threshold co~"pare 308. The peak amplitude of the full-bandwidth input signal at time n is ~cum~d to be the sum of these peak amplitude estim~t~s at time n, or P7~n) = PL(n) + PH(n) where PL(n) = held peak-amplitude ç~tim~tP for low frequencies at time n, and PH(n) = held peak-amplitude estim~te for high frequencies at time n.
For this particular application, it is ~sumed that only the high frequencies contain PLI, the~fo,t;, gain reduction should be applied only to these frequencies.
Whenever P7~n) el~ce~lc the reference level, gain select 310 establishes a gain factor g(n) to apply to the high frequencies in order to reduce the full-bandwidth peak amplitude to substantially the reference level. This gain factor may be obtained by solving the following two indep~ndent linear equations in two unknowns. These equations are eAI)~ssed in terms of peak-amplitude estim~tes which are norm~li7çd or scaled by the desired reference level such that the norm~li7e~ desired peak amplitude is equal to one; thus:
PL(n) + ph~n) = pT~n) pL(n) + g(n) pH(n) = 1.0 where pL(n) = norm~li7ed low frequency held peak-amplitude çstim~te at time n, pN(n) = norm~li7~d high frequency held peak-amplitude estim~te at time n, p7(n) = norm~li7~d full-bandwidth signal peak-amplitude estim~t~ at time n, and g(n) = gain factor required to optimally correct PLI at time n.
By solving these two equations simultaneously, the gain factor is seen to be Wo 94/19883 21 5 ~ 8 8 3 PCT/US94/01639 g(n) = I _ p7l~n)-l l-pL(n) pH(n) pH(n) (3) The gain table discu~ed above can be used to provide a gain factor which b~l~nces PLI
reduction against loss of ap~)arent loudness in the output signal. Conceptually, gain factor g(n) is the ratio between the high-frequency signal's "proper" level and its peak amplitude. This 5 ratio can be ~xpressed as 1/(1 +e) where e is the normalized deviation between the norm~li7ed high frequency signal peak amplitude p~(n) and the norm~li7ed amplitude required to correct PLI. Referring to equation (2), the index i into gain table T[i] is seen to be the actual deviation between the actual peak-amplitude estim~te x(n) and the actual reference level A,~,.
The actual deviation may be obtained by scaling the normalized deviation e with the actual 10 reference level A~". The.efore, gain select 310 may obtain a gain factor from the gain table entry TleA~,J.
Although the gain select function directly affects the ~,rol~l,ance of a particular embodiment, it should be understood that no particular gain select function is critical to the practice of the present invention. Other duplex gain select functions are possible and will be 15 ap~arent to those with oldinaly skill in the art.
e. Control Filter The dynamic characteristics of the control system must be established by b~l~ncing two conflicting goals. On the one hand, the control system should respond quickly to the onset of PLI and recover quickly after PLI has stopped so that the a~ale,-t loudness of the reproduced 20 signal is not affected more than nPces~ry. On the other hand, the conkol system should not respond and recover so quickly that it generates audible modulation distortion in the reproduced signal. The particular embodiment shown in Figure 3 from input path 302 up through gain select 310 allows the control system to respond and recover as quickly as possible. Control filter 312, however, restricts the rate of response to the generation of 25 audible artifacts.
Referring to Figure 1, gain element 114 responds to a gain control signal received from control system 122 to amplitude modulate the signal arriving along path 112. As a result of the arnplitude modul~tion, distortion in the form of ~ideb~n~c is generated.
Control filter 312 shown in Figure 3 is essentially a low-pass filter de~ign~ to limit 30 the frequency of the amplitude modulation and thereby limit the bandwidth of the sideb~nds.
The LPF bandwidth of control filter 312 should be chosen so that the sideband bandwidths are confined to within a psychoacoustic critical bandwidth of the spectral cG"~nents mocl~ ted wo 94/1g883 1 5 ~ 8 8 ~ PCT/US94/01639 by the variable gain element. In the preferred embodiment for the FM-broadcast STL
application (li~cllssed above, the function is implemented as a third order IIR filter with a cutoff frequency of 880 Hz, a stopband attenuation of -40 dB, and 0.09 dB of ripple in the low-pass band.
By ex~mining the effects of the modulation sidebands with respect to the psychoacoustic m~c1~ing thresholds of the modul~ted signal, the characteristics of the control filter may be specified so as to permit the fastest control system response possible consistent with psychoacoustic m~ ing of the modulation artifacts.
C. Sl~ lur~ of Alternate EmboJi~ s Many variations in embodiments of the present invention are possible. The following describes dirrerences between the alternate embodiments and the basic structure described above.
One embodiment of control system 122 is shown in Figure 6. Referring to Figure 6, 604 esli,n~s the peak ~mplitude of the input signal received from path 602 and passes the estim~te to peak hold 606. Peak hold 606 passes to threshold co~nl)a~e 608 a pulse-shaped signal which holds the estim~ted peak amplitude for a specific duration. Threshold colllpalc 608 colllpalcs the ~mplitude of the peak hold signal with a reference level and passes the results of the colllpdlison to gain select 610. Gain select 610 selects a gain factor in response to the signals received from threshold COlllp~UC 608 and passes the sele~ted gain factor to control filter 612. Control filter 612 ge.-e,dtes along path 614 a gain control signal in response to the s~ ted gain factor.
The functiQnc ~ro.llled by peak estim~t~r 604, peak hold 606 and threshold COIllpalc 608 are 5~5~ 1ly similar to the functions described above for the collc~nding el~ment~
shown in Figure 3 along the path carrying the first frequency subband con,ponent. Unlike the cou"te~ threshold colllpare 308, however, threshold col,lpa,c 608 receives an estim~te of the full-bandwidth input signal directly from one peak hold signal. Furthermore, threshold co,llparc 608 need not pass any peak hold signal to gain select 610, but can pass only the difference between the peak hold signal and the reference level.
Gain select 610 establishes a gain factor in response to the output of threshold co,llpalc 608. If the peak-amplitude estim~te is less than the reference level, a gain factor equal to one is select~. If the peak-amplitude e~l;",~te exceeds the reference level, however, a gain factor is select~ which reduces the amplitude of the frequency subband colllponent cont~ining PLI.
~_ 17 2154883 The gain select function may be implemented using equation (1) above, by selecting a gain factor equal to the ratio between the reference level and the peak-amplitude estim~e, or by using a gain table such as that shown in equation (2). Although the choice of gain select function may greatly affect system pe~ro~ aulce, no particular function is critical to the 5 practice of the present invention.
It should be appreciated that gain select 610 uses a simplex gain select function which generates a gain control signal in response to the full-bandwidth input signal. Other embodiments inco,~,d~ g duplex gain select functions such as those described above may generate a more optimal gain factor by responding to signal levels in both the full-bandwidth 10 and in the portion of the bandwidth in which PLI is to be limited. Although the implcm~nt~tion shown in Figure 6 may be suboptimal, it provides acceptable performance for a variety of applications including the broadcast STL application di~cucc~od above, and it ,~ui,es fewer proceccing resources than those required by the more optimum duplex embo-iiments.
Figure 7 illustrates an alternative but equivalent embodiment to that shown in Figure 1.
As shown in Figure 7, the splitter is imple--,æl-~ed with a low-pass filter and a subtractor, and the combiner is implement~ with an adder. LPF 708 receives the delayed full-bandwidth signal from path 706 and passes the low-frequency spectrum along path 710 to subtractor 711 and combiner 718. Subtractor 711 receives the delayed full-bandwidth signal from path 706, 20 subtracts the low-frequency col"ponents from the full-bandwidth co",~nents and passes the resl-lting signal having only high-frequency co"lponents along path 712 to gain elemt nt 714.
Gain elPm~nt 714 receives a gain control signal from control system 722 along path 724, applies a variable gain factor to the signal received from path 712, and passes the result to CGIllbiner 718. Colllbi~er 718 adds the signals received from paths 710 and 716 to generate 25 a full-bandwidth output signal along path 720.
Figure 8 illustrates an alternative embodiment to that shown in Figure 1 for imple...~ l;onc inco,~ldting a duplex gain select function. The functions pe,rol",ed by splitter 804 and delay elements 808 and 814 are interchanged as collll)~cd to that shown in Figure 1. As a result, control system 826 does not require a sc;pal~te splitter. The gain select 30 function can directly apply equation (3), ~liccucced above, to the first and second frequency subband co"")onents received from paths 812 and 806, res~e~;lively.
Figure 9 illustrates an alternative embodiment using a hybrid limiter structure similar to that shown in Figure 10c. Delay element 904 receives a full-bandwidth signal from path 902. Splitter 906 splits the delayed full-bandwidth signal into a first frequency subband WO 94/19883 21 S ~ 8 ~ 3 PCT/US94/01639 co-,-ponent, passed along path 910, and a second frequency subband component passed along path 908. Gain element 912 receives the first frequency subband co.n~nent from path 910, applies an adaptive gain factor to the first frequency subband co---pol-el1t and passes the result along path 914 to combiner 916. Combiner 916 combines the first and second frequency S subband co---ponents received from paths 914 and 908, respectively, and generates an output signal along path 918. Splitter 920 splits the fuli-bandwidth signal received from path 902 into a first frequency subband co---ponent, passed along path 922, and a second frequency subband co.-.~nent passed along path 928. Gain element 924 applies an adaptive gain factor to the first frequency subband co,.-ponent and passes the result to combiner 930. Combiner 930 10 combines the first and second frequency subband components received from paths 926 and 928, l~s~ ely, and generates a quasi-output signal along path 932. Control system 934 receives the quasi-output signal from path 932 and passes a gain control signal along path 936 to gain elements 912 and 924.
~ use output-controlled limiters are incpnsitive to precise control system 15 characteristics, embo~imentc of the present invention according to Figure 9 do not require accurate gain tables such as that .liccllcspd above.
The present invention may also be applied to multi-ch~nnPl applications. In one digital embo~limPnt incol~ld~ing a control system in accordal~ce with the structure shown in Figure 6, the maximum peak-amplitude estim~te across all channels is ~rtcented to the input 20 of a single peak hold 606. The rennlinder of the control system ~lrOl...s in a manner s,ll.sl.n~i~lly the same as that described above.
In an alternative embodiment which requires fewer plocPcc;ng resources, the largest m~gnitude sample for all ch~nn~plc is passed along path 602 to a single peak estim~tor 604.
Peak estimator 604 interpolates these largest m~nitude samples and passes a peak-amplitude 25 esli.n~te to peak hold 606.
It will be appar~;--t that the various functions in the control system structure as shown in Figure 6 may be shared across multiple ch~nnPlc in a similar manner as desired.
Many other control system embodiments are possible for multi-ch~nnel applications;
however, each must balance reduction in appalent loudness of each rh~nnçl, distortion of 30 relative ch~nnçl loudness, and modulation of frequency subband co..-ponents contdining PLI
by other frequency subband co~--ponents without PLI. For example, the maximum PLI-to-reference level ratio across all ch~nnels may be used to select a gain factor for each ch~nnel using the duplex gain select function described above. In this case, the duplex gain select wo 94/19883 215 4 8 8 3 PCT/US94/01639 function is applied using the peak-amplitude estim~tçs for each respective channel in combination with the common maximum PLI ratio.
The alternatives described above are given by way of example only, and illustrate that the present invention is applicable to a broad variety of structures and implementations.
DIGITAL AUDIO LIMITER
Technical Field The invention relates in general to limiting the peak amplitude of an audio signal.
5 More particularly, the invention relates to limiting the peak amplitude of one or more frequency subbands of an audio signal while preserving the apl)afe-lt loudness.
B~l k~round Art There is considerable interest among those in the fields of bro~-lc~cting and recording to reduce the amount of information required to transmit or record an audio signal intendeA
10 for human perccpLion without degrading its subjective quality. Analog signals with reduced informational require",ent~ can be carried within narrower bandwidths, and digital signals with reduced information requilc,,,cnts can be carried at lower bit-rates.
One co"""on technique used to reduce the info""ational ~cquil~ ents reduces the dynamic range of the audio signal to be transmitted or recorded. Dynamic range control is 15 used to protect euuip,,,. nl from excessively high-amplitude signals and to achieve certain artistic results. In general, the overall goal of dynamic range control is to alter the dynamic range of an audio signal without introducing any other ~rcepLible distortion.
In bro~ç~cting, for example, the dynamic range of audio signals is controlled prior to broadcast to avoid overloading tr~ncmiccion equipment and/or to avoid severe audible 20 distortion. Similar concel"s apply to conventional tape and disc rccofdillgs. The control of dynamic range may be accomplished by automatic techniques such as "CO"~pf ssion" and "limiting" or by appropfiate manual gain setting~c.
Dynamic range "col~ples~ion" reduces the dynamic range of an input signal by applying a variable gain factor varied in response to input signal amplitude. A "co",p~ssor"
25 provides for signal amplitude dynamic range co"~prcssion A "limiter~ provides "limiting"
which is a special case of co"~plession, preventing the peak-amplitude of an input signal from -Yce~Aing a specified level by applying a very low gain factor to high-level signals.
A second common technique used to reduce the informational l~uire~ en~s of an audio signal reduces the amount of information used to represent or code the audio signal; however, 30 as the amount of information is reduced, encoding inaccuracies increase and may become audible in the form of "coding noise." Coding noise degrades the subjective quality of the coded signal. So-called psychoacoustic-coding or pclccplllal-coding techniques attempt to reduce the informational requirements of an audio signal without introducing audible coding noise. Two examples of "split-band" pclceplllal-coding techniques are subband coding and transform coding. Perceptual-coding techniques exploit a characteristic of human hearing; a stronger signal may mask or render inaudible a weaker signal if the two signals are sufficiently 5 close in frequency. By splitting an audio signal into narrow frequency bands and independently coding the signal energy in each band, the aural effect of the coding noise is more likely to be inaudible because it is confined to the same frequency band as the coded spectral energy.
Coding systems which implement ~rce~ut~lal-coding techniques attempt to reproduce 10 a ~p~ ;on of the input audio signal which preserves the perceived loudness of input signal spectral co~pon~ . This is often accomplished by preserving some measure of spectral amplitude such as root-mean-square (RMS); however, many ~r~plual-coding systems lead to uncertainties in the peak-amplitude level of the reproduced signal. These uncertainties may include increases in peak-amplitude, I~;îel.ed to herein as "peak-level increase" or PLI, 15 which is tolerable in many coding applications where it is in~u~ible The Studio-Tr~ncmitter Link (STL) for bro~lc-~cting~ which delivers an audio signal origin~ting from a studio to a broadcast tr~ncmitter~ is one example of an application which cannot tolerate excessive amounts of PLI. In one embodiment, an STL includes a colnplel,sor and limiter which reduce the dynamic range and limit the peak-amplitude level of an audio 20 signal, a ~r~eptual-encoder which reduces the informational capacity requirements of the audio signal, a communications ch~nntol for delivering the encoded signal, and a p lcepl~lal-decoder for reprodllçing the co,np~ss~ and limited audio signal for subsequent bro~ ct~
ReC~u-ce of PLI, the peak-amplitude of the signal reproduced by the pe~eptllal-decoder may sometimeS exceed the capabilities of the broadcast tr~ncmittPr even though the peak amplitude 25 of the audio signal input to the pel~e~ual-encoder is pfoyelly limited. Tr~nsmitter overload resulting from PLI can create audible distortion and/or impermissible broadcast conditions such as excessive FM deviation.
Known techniques for controlling PLI include clippels, inct~nt~n~us gain reduction amplifiers, and conventional wide-band limiters. Unfortunately, these techniques introduce 30 undesirable audible distortion in the reproduced signal. Clippers generate excessive harmonic distortion. Inct~nt~ne~us gain reduction amplifiers, in effect, smear spectral colllponents in the frequency domain. Conventional wide-band limiters reduce the perceived loudness of the reploduced signal.
wo 94/19883 215 ~ ~ 8 3 PCT/US94/01639 Disclosure of Invention It is an object of the present invention to provide for a signal ploce~ing system in which both al)pa ent loudness of the reproduced audio signal and PLI are controlled within acceptable levels.
It is a further object of the present invention to provide for a signal processing system in which a high-quality leprese,-td~-on of the input audio signal is reproduced. These objects are achieved by the invention as claimed.
In accorddnce with the te~^hings of various aspects of the present invention in one embodiment, a signal pr~ceccing system receives a peak-amplitude limited input audio signal, generates a plocessed audio signal in response to the input audio signal such that peak-level increase may be present, estim~tes the peak-level increase of the full-bandwidth processed audio signal, and generates an output audio signal by applying to the portion of the full-bandwidth subject to PLI a gain factor adapted in response to the estim~tçd peak amplitude.
In accoldance with the tç~ ingc of another aspect of the present invention, a signal proceccing system receives a peak-amplitude limited input audio signal, generates a processed audio signal in response to the input audio signal such that peak-level increase may be present, estim~tçs the peak-level increase of the full-bandwidth processed audio signal, and gen~,dtes an output audio signal by applying to portions of the full-bandwidth subject to PLI a plurality of gain factors adapted in response to the estinl~tçd peak amplitude.
The present invention is generally applicable to signal proceccing which is subject to PLI. As ~iiscussed above, many p~,~;t;plual-coding systems and methods are known in the art which gel-cldte, in ~Jonse to a peak-amplitude limited input audio signal, a plOCf ~ed audio signal with PLI. Of cignifi~nce to practical embodiments of the present invention, the inventors have determined empirically that in broadcast STL applications, the predo."inant audible effects caused by PLI generated by pe~cepl~lal-coders are confined to high-frequencies above about 5 kHz. It is therefore possible to adequately correct for PLI by applying an appl~pliale gain factor to only those high-frequency signal components.
A low-pass filter (LPF) with a frequency-dependent phase shift is another example of signal processing which can preserve app~G--t loudness but permit PLI. For example, such 30 a 20 kHz LPF receiving a 6 kHz square wave input signal will pass the third harmonic at 18 kHz with its phase shifted relative to the phase of the fund~mçnt~l frequency. If the relative phase shift is of an a~o~liate amount, the third harmonic will combine with the fun-l~m~o-n~l frequency to increase the peak amplitude rather than to flatten the waveform and reduce the 215~88~ 4 peak amplitude. Although the a~arent loudness of the low-pass filtered signal may be preserved, the peak-~mplitude of the passed signal will increase.
The present invention is also applicable to practical implementations of co",ponents such as digital-to-analog converters and anti-aliasing filters which generate PLI. Other applications will be known to those skilled in the art.
Various aspects of the present invention and its plefelled embodiments are set forth in greater detail in the following "Modes for Carrying Out the Invention" and in the acco",panying drawings. It should be appreciated that the following liccll~cion sets forth several embo lim~nt~ by way of example only, and that these examples are not intended to ~resent any limitations in application or implementation.
Although the following ~is~uccion is more particularly dir~c~ed toward digital split-band coding within an STL application, the present invention is not so limited. As discu~ briefly above, aspects of the present invention are applicable to a broader variety of signal proce~sing systems. Furth~.",~"t;, the present invention may be embodied in systems implemented by analog techniques as well as digital techniques.
Brief D~ t.on of Drawin~c Figure 1 is a functional block diagram illustrating the structure of one embo iimçnt of an audio signal limiter according to various aspects of the present invention.
Figure 2 is a functional block diagld", of an STL incorporating an embodiment of the present invention.
Figure 3 is a functional block diagram of one embodiment of a control system according to various aspects of the present invention.
Figure 4 is a functional block diagram of one embodiment of a peak-amplitude estim~tor.
Figure S is a hypothetical graphical re~ sen~ation of one peak hold function.
Figure 6 is a functional block diagram of one embodiment of a control system according to various aspects of the present invention.
Figure 7 is a functional block diagram illustrating an alternative band-splitting structure within an embodiment of an audio signal limiter according to various aspects of the present invention.
Figure 8 is a functional block diagram illustrating the structure of one embodiment of an audio signal limiter according to various aspects of the present invention.
wo 94/19883 215 4 8 8 3 PCT/US94/01639 _ 5 Figure 9 is a functional block diagram illustrating an alternative hybrid input-controlled/output-controlled structure according to various aspects of the present invention.
Figures 10a-lOc illustrate functional block diagrams of three types of limiters.
Modes for Carryin~ Out the Invention S A. Intro~-c~ic~l Figures 10a-lOc illustrate functional block diagrams of three types of limiters. The limiter shown in Figure 10a is an output-controlled limiter which uses negative fee~ibacl~.
Gain element 1004 receives an input signal from path 1002 and generates an output signal along path 1006 by applying a variable gain factor to the input signal. Control system 1008 receives the output signal from path 1006 and passes a gain control signal along path 1010 which selects the gain factor used by gain element 1004. Output-controlled limiters are very tolerant of variations in the ch~dclelistics of the control system and the gain elem~-nt~ but they cannot avoid ovelsl-oot in the output signal because of the inherent delay in the control system.
The limiter shown in Figure 10b is an input-controlled limiter. Delay element 1024 receives an input signal from path 1022 and passes the delayed signal along path 1026 to gain element 1028. Control system 1032 receives the input signal from path 1022 and geneldles a gain control signal along path 1034 which selects the gain factor used by gain elemçnt 1028.
Gain element 1028 receives the delayed input signal from path 1026 and gene.~dtes an output signal along path 1030 by applying a variable gain factor to the delayed input signal. Input-controlled limiters require a control system and a gain element with precisely known char~ct~ri~tics. This is usually much easier to implement with digital technologies than it is to implemçnt with analog technologies; however, input-controlled limiters can avoid overshoot in the output signal by using a delay elen ent The limiter shown in Figure 10c is a hybrid of the other two types of limiters. Delay elçmçnt 1044 receives an input signal from path 1042 and passes the delayed signal along path 1046 to gain element 1048. Gain element 1052 receives the input signal from path 1042 and geneldtes a quasi-output signal along path 1054 by applying a variable gain factor to the input signal. Control system 1056 receives the ~uasi-output signal from path 1054 and gene,~es a gain control signal along path 1058 which selects the gain factor used by gain element 1052 and gain element 1048. Gain element 1048 receives the delayed input signal from path 1046 and g~ t~s an output signal along path 1050 by applying a variable gain factor to the delayed input signal. The hybrid limiter does not require a control system and gain elements with precisely known characteristics, but it does require two gain elements with precisely WO 94/1s883 PCTIUS94/01639 matched characteristics. This is usually much easier to implement, even with analog technologies. Furthermore, by using the delay element, the hybrid limiter is also able to avoid overshoot in the output signal.
Figure 2 is a functional block diagram of a portion of a broadcast system including an S STL incoll~o,~ling one embodiment of the present invention. The following discussion more particularly describes various embodiments of the present invention for this application. The embodiments described below incorporate an input-controlled~limiter, but it will be readily a~)pa,~ nt in light of the prior art that these different embQdiments may be mo lified to also incorporate an output-controlled limiter or a hybrid limiter as described above.Referring to Figure 2, co~ )ressor/limiter 204 receives a studio signal from path 202 and generates a peak-amplitude limited re~ sent~tion of the studio signal which it passes to encoder 206. Encoder 206 generates an encoded signal which it passes along communication path 208. Decoder 210 receives the encode~ signal from communication path 208, reproduces the colllp.essed and limited signal, and passes the reproduced signal to limiter 212. Limiter 212 co~,~cls any PLI in the reproduced signal and passes the col,ecled signal along path 214 to a broadcast tr~ncmitter. It should be understood that co,llpl~ssor/limiter 204 does not form part of the present invention and is not required to practice the present invention. For example, app,~iate manual gain settings alone may be sufficient to bound signal dynamic range; however, collll)~ssion is sometimes used to reduce dynamic range and limiting is sometimes used to prevent ac~idPnt~l violations of prescribed peak amplitude limits.
For ease of ~1iscu~ion~ pr~re.led digital embodiments of the present invention for an FM-broadcast STL application are desclibed which assume that the reproduced audio signal delivered to limiter 212 has a 15 kHz bandwidth and is sampled at a rate of 44.1 kilo-samples/sec. Changes required to implement an embodiment of the present invention for use with signals having different characteristics are within the abilities of those having ordinaly skill in the art.
B. Basic Structure Figure 1 illustrates the structure of one embodiment of an input-controlled limiter according to various aspects of the present invention. Delay element 104 receives an input signal from path 102 and passes the delayed signal along path 106 to splitter 108. Splitter 108 splits the delayed signal into at least two sub-signals, passing a first sub-signal along path 112 to gain element 114 and passing the rem~ining one or more sub-signals along path 110 to colllbiner 118. Control system 122 receives the input signal from path 102 and generates a W O 94/19883 215 4 8 8 3 PCTrUS94/01639 _ - 7 -gain control signal along path 124 which selects the gain factor used by gain element 114.
Gain element 114 receives the first sub-signal from path 112 and generates a second sub-signal along path 116 by applying a variable gain factor to the first sub-signal. Combiner 118 generates an output signal along path 120 by combining the sub-signals received from paths 1 10 and 1 16.
Splitter 108 divides the delayed input signal into two or more sub-signals leplcsenting two frequency subband col"ponents. The first frequency subband component is subject to PLI. The second frequency subband colnponent includes that portion of the s~ , Lllll in which PLI does not exist. The first frequency subband component which may have PLI is ~ple~nted in Figure 1 by the first sub-signal passed along path 112 to gain element 114. The other frequency subband component is lcpresented in Figure 1 by the one or more sub-signals passed along path 110 to colllbiner 118.
1. Signal Pr~ceccing Path The structure shown in Figure 1 has two paths. A signal procesc;ng path includesdelay 104, splitter 108, gain element 114, and combiner 118. A se~,a.dte control path, below, includes control system 122.
Delay elernent 104 may be implemented by any technique applupliate for delaying the audio signal received from path 102. The duration of the delay is usually set subsl~n~ y equal to the length of time required for control system 122 to respond to PLI requiring 20 correction. One method which may be used to establish this duration is to establish the delay which yields the maximum cross-correlation score between the delayed signal on path 112 cont~ining PLI and one minus the gain factor (1-g) generated by control system 122.
Splitter 108 may be implemented by any technique for dividing a signal into frequency subbands including, but not limited to, analog filters, digital filters, and so-called frequency-25 domain transforms.
Combiner 118 may be implemented by any technique for combining the sub-signals into a full-bandwidth signal. The combiner implementation technique will normally be the inverse of the implement~tion technique chosen for splitter 108.
Gain element 114 may be implemented by any technique a~r~liate for the signal 30 received from path 112. For Py~mple, operational amplifiers may be used in analog systems and m-lltipli~tion or scaling may be used in digital systems. The resolution and range of the gain element should be s~lected in conjunction with the operational characteristics of control system 122, discussed below.
WO 94/19883 215 ~ 8 8 3 PCT/US94/01639 2. Control Path One embodiment of control system 122 is shown in Figure 3. Referring to Figure 3, splitter 303 splits the input signal received from path 302 into two frequency subband co",~nents in the same manner as that performed by splitter 108 discussed above. The second frequency subband component, which is a~cumed not to contain PLI, is passed to peak estim~tor 305. Peak estim~tor 305 estim~t-Ps the peak ampl~itude of the second frequency subband component and passes the estim~te to peak hol~ 307. Peak hold 307 passes to threshold co",l,~rc 308 a pulse-shaped signal which holds the estim~t~d peak amplitude for a specific period of time.
The first frequency subband c(jmponent, which may contain PLI, is passed to peakestim~tor 304. Peak Pstim~tor 304 estimates the peak amplitude of the first frequency subband co"~ponent and passes the ectim~t~p to peak hold 306. Peak hold 306 passes to threshold CG~pale 308 a pulse-shaped signal which holds the estim~ted peak amplitude for a specific period of time.
Threshold col"palc 308 obtains a peak-amplitude estim~te of the full-bandwidth input signal by col"bining the two peak hold signals and cG,npales the e~ te with a reference level. Threshold co"")are 308 passes the two peak hold signals and the results of the cGI"pdlison to gain select 310. Gain select 310 selects a gain factor in les~n~ to the signals received from threshold CGIllparC 308 and passes the sPle~ted gain factor to control filter 312.
Control filter 312 generates along path 314 a gain control signal in respon~ to the select~P~
gain factor. The functions rcp-~3ented in Figure 3 affect both the steady-state and the dynamic chalac~llstics of the control system.
The following rliscn~ion of the embodiment shown in Figure 3 is directed toward digital implementations; however, it should be applcciated that COIl~ s~onding functions may be implP-m~nted using analog techniques.
a. Peak F.ctim~tor In a plcrelled embodiment of the present invention for an FM-broadcast STL, the first frequency subband col~pol-ent contains high-frequency spectral co,n~nent~ up to about 15 kHz. Peak esli...~tor 304 uses an "upsampling" filter to interpolate the samples of the first 30 frequency subband component, thereby improving the accuracy of the peak-amplitude estim~t~p.
~lthough samples taken at or above the Nyquist rate permit accurate recovery of a sampled signal, it is well known that the samples themselves do not accurately r~le3ent the signal peak amplitude. In many digital applications the only conceln is the ability of the digital system to process and transmit the samples themselves. In applications such as a broadcast STL, wo 94/19883 215 4 8 8 3 PCT/US94/01639 _ g however, the actual peak amplitude of the analog signal reproduced from the digital signal is of concern. There~ore, the peak estim~tor function interpolates between samples to obtain a more accurate çstim~te.
In the same p~cfelred embodiment, the second frequency subband component contains 5 only frequencies below about 5 kHz. An accurate peak amplitude can be estimated accurately from samples taken at a rate of 44.1 kHz; therefore, peak estim~tor 305 does not need to interpolate the samples of the second frequency subband component.
Figure 4 shows a functional block diagram of a preferred embodiment of an interpolating peak estimator. Upsampler 404 and upsampler 406 are each two-times10 upsampling filters, and together the two upsamplers generate four samples in response to each input signal sample received from path 402. Each u~ --pler can be implem~nt~l efficiently as a low-pass half-band FIR filter. More particularly, for the FM-broadcast STL application, filter characteristics are chosen so as to obtain a 15 kHz p~sb~nd with -40 dB of attenuation in the stopband The filter for upsampler 404 comprises seventeen taps and upsamples to a rate of 88.2 kilo-samples/sec. The filter for upsampler 406 comprises nine taps and upsamples to a rate of 176.4 kilo-samples/sec.
Selector 408 co,n~ares the four most recent interpolated samples and selects the one with the largest absolute value. Selector 410 coml)~es the magnitude selected by selector 408 with a stored value. Selector 410 passes the larger of the two values along path 412 and saves the value s~lected by selector 408 as the new stored value.
Conceptually, the peak estim~tor shown in Figure 4 up~rnples the input signal by four times and, for each set of four interpolated samples, selects the largest absolute value from the eight most recent interpolated samples.
b. Peak Hold The functions r~plesented by peak hold 306 and peak hold 307 shown in Figure 3 are nPcçsc~ry because the limiting action of control system 122 and gain element 114 shown in Figure 1 is not instantaneous. In order to minimi7~ the audible effects of the limiting action within the signal procescing path, it is illl~ t to control the rate at which the gain factor applied by gain element 114 is changed. Various considerations are discussed below. It is 30 sufficient at this point to realize that the gain factor should not be changed in~t~nt~neously.
Therefore, unless some col-,pensation is made, the control system will not be able to adequately respond to short-duration PLI intervals. The function r~l senled by peak hold 306 and peak hold 307 in Figure 3 is one way in which compensation may be made.
Wo 94/19883 215 ~ 8 8 3 PCT/US94/OlG39 Each peak hold function generates a pulse signal of duration H having an amplitude equal to the peak-amplitude estim~t~- received from the peak estim~tors. For example, in response to a peak-amplitude estim~te of xl at time t" the peak hold function generates a pulse signal with an amplitude equal to xl, starting at time t, and ending at time tl +H. If a larger S peak-amplitude estim~te x2 is received at time t2 before the first pulse ends, however, the peak hold function imme~i~tely generates another pulse signal with an amplitude equal to x2 of duration H starting at time t2 and ending at time t2 +H. If a ~eak-amplitude estim~te X3 smaller than x2 is received during the current pulse signal at time~t3, the peak hold function generates a pulse signal with amplitude equal to X3 starting at the end of the current pulse signal and 10 ending at time t3+H. Figure 5 provides a hypothetical graphical ~ esentation of this function with a hold duration of 4T.
The following pseudo-code program seg.,l.~nt illustrates the logic of an embodiment which implements the peak hold function described above and illustrated in Figure 5.
i--O
Y--O
X(O:H)--O
while TRUE
X(i)--x(n) if Y ~ X(i+1) then Y--ms~(Y,X(i)) else Y--max(X(I~,,X(i+H-l)) endif y(n)--Y
i--i+l endwhile where x(n) = peak-~mplitude estim~t~- at time n, H = the peak hold duration, and y(n) = peak hold function output at time n.
The program segment implements a circular buffer X of length H; therefore it should be understood that the variable i is an index or pointer to the buffer and the notation i+k r~,csenls the eA~,- ssion (i+k) - int( i k~ H
where int(q) = the integer portion of q.
After initi~li7ing the variables i, Y and circular buffer X, the program se.gment contained within the "while" group is r~peLili~rely executed. As each peak-amplitude estim~te wo 94/1g883 215 4 8 8 3 PCT/US94/01639 x(n) at time n is received, its magnitude is stored in circular buffer X. The current value of Yis co,-,palcd to the value stored H time periods earlier in buffer X. If the two are not equal, Y is set equal to the larger of the magnitude just stored in buffer X or the current value of Y.
If they are equal, this indicates that the peak value has been held for duration H and Y is set equal to the largest of the last H magnitudes stored in buffer X. Many other implementations are possible.
The length of duration H is a design choice which should be made together with other design choices affecting the gain select and the control filter functions, discussed in more detail below.
c. Threshold Compare Threshold co-~palc 308 co,npa,es an çstim~te of the full-bandwidth input signal peak amplitude with a reference level and passes the results of the comparison to gain select 310.
In an embodiment according to Figure 3, an estim~tç of the full-bandwidth input signal may be obtained by combining the peak hold signals from peak hold 306 and peak hold 307.
In a p~cr~llcd embodiment of the present invention for broadcast STL applications such as that shown in Figure 2, the reference level is pre~efined according to the peak-amplitude limit i~.pose~ by cG",pl(ssor/limiter 204 upon the studio signal. The threshold co,np~e reference level is the desired peak-amplitude limit of the signal reproduced by pe,eept~al-decoder 210.
In principle, the r~f~ence level need not be predefined but could be provided by an external signal. The desired peak-amplitude level could be passed with the signal having PLI.
Referring to Figure 2, for example, the desired peak-amplitude limit could be encoded by ~ncodçr 206, passed with the encoded signal along path 208, extracted by de~oclçr 210 and provided to limiter 212. Many variations are possible.
d. Gain Select Gain select 310 establishes a gain factor in response to the output of threshold co"~pa,e 308. If the peak-amplitude estim~te of the full-bandwidth signal is less than the reference level, a gain factor equal to one is selected. If the peak-amplitude estim~tç ~xcee~ the reference level, however, a gain factor is selected which effectively reduces the amplitude of the frequency subband co",ponent con~ining PLI such that the peak amplitude of the output signal is reduced to the desired amplitude defined by the reference level. Additional details ~,~ining to the design of limiters may be found in Bosi, "Low-Cost/High-Quality Digital Dynamic Range Processor," AES 91st Convention, New York, October 1991, Preprint 3133, which is incol~o,~ted by reference in its entirety.
WO 94/lg883 21 S 4 8 8 3 PCT/US94/01639 For conventional full-bandwidth limiters~ a gain factor g may be established at time n subst~nti~lly in accordance with [7h - 20 bg ~(n)~
where g(n) = selected gain factor at time n, S Th = reference level e~p,lcssed in dB, x(n) = peak-amplitude estim~tt~ at time n, and R = cG,~plession ratio. .; ~ `
Alternatively, a gain factor g(n) may be obtained from the ratio between the reference level Th and the peak-amplitude estim~te x(n). This is equivalent to the results obtained from 10 equation (1) where the co~ ,rcs~,ion ratio R is infinitely large.
By using equation (1) and setting the co"~p,ession ratio R at some finite value, gain select 310 is able to select a gain factor which balances PLI reduction with the decrease in appar~l loullnrss of the output signal. For many broadcast STL applications, a cc""~,lession ratio R=100 effectively removes PLI.
For ~,u~ oses of the present invention, however, the gain factor c~lcul~ted acco.~ing to equation (1) may not provide nearly enough limitjng if equation (1) is applied to only a single peak-amplitude estim~t~. At least two peak-amplitude estim~t~s must be used in applications such as the broadcast STL iiccllc~d above where PLI is confined to certain frequency bands which do not rep,esent the p~edol..ill~nt spectral energy for the full-bandwidth signal. Neve. Iheless, it is useful to first describe a simpler gain select function which ~ccllmes that the frequency band cont~ining PLI l~iesents subst~nti~lly the full-bandwidth signal energy. For ease of ~liccllcsion~ this gain select function is ~felred to herein as "simplex gain select" because it requires only one input source, the çstim~ted peak amplitude of the frequency subband cG-~-ponent containing PLI.
i) simplex gain select Although equation (1) defines the desired gain to select, a straight-forward calculation is computationally intensive. In a pç~felled embodiment, a more effici~nt process is implçmçnt~ by selecting the desired gain from a pre-computed gain table. The gain table T
contains a set of gain factors which may be defined by the following pseudo-code program sçgm~nt WO 94/19883 215 ~ 8 8 3 PCT/US94/01639 __ - 13 Th---2 R--lOO
A",p--16 S S--(R-l) I R
Am~ ~ 2 -l A ~ lo~/20 A
Tsz _ (A",~" - A,b) / Ad,p ~--O
A--A,b for i from O to Tsz l T~l--10~-~/20 A . A + A"~p ~--Th -201cgA
endfor where N = number of bits available for digital lcplcscntation, Th = reference level in dB, R = CGIllp~ Sion ratio, A~ep = amplitude step siæ, A"",~ = maximum digital le~fe~n~;1tion, A,h = absolute reference level, Tsz = gain table size, and T[il = gain factor i in the gain table.
In one embo~iment sixteen bits are used to express signal amplitudes, Ihe1cforc, N is set equal to l6. It should be appleciatcd that the gain table T contains anti-log gain factors, the~cfo~c the gain increment a between ~ cent table entries is not constant. The amplitude step size A~ep is selected such that the largest gain increment ~\ between ~ Cçnt table entries is not too large. In a p1cfe11~ emb~im~-nt, A~ep is set to sixteen which ensures that /~ is always less than 0.05 dB.
In the ylcfe11~ embodiment, gain select 3lO receives from threshold co",parc 308 a signal in(~ ting the difference D between a peak-amplitude estim~te y(n) and the 1efelcnce level A,h. Gain select 3 lO uses D as an index into the gain table T and thereby selects the gain factor g al,pro~liate to reduce the PLI causing the pcak amplitude error, or g(n) = Tly(n)-A~h~. (2) Many other implementations of the gain select function are possible.
ii) duplex gain select In applications having PLI oc-;u11ing in frequency subband co",l,onc1-ts which have significantly less than the full-bandwidth spcctral energy, simplex gain select does not provide enough limiting to optimally reduce PLI to desired levels. A more optimum gain select function, r~fe"~ to herein as "duplex gain select," requires two peak-amplitude estim~tps.
The conceptual basis for duplex gain select assumes that the sum of the peak amplitudes in various frequency subbands equals the peak amplitude of the full-bandwidth input signal.
For example, two bandwidth limited signals having peak-to-peak amplitudes of 0.4 volts and 0.6 volts, respectively, when combined, will form a signal having a peak-to-peak amplitude of one volt. Although this assumption is not precisely correct, it is accurate enough for the purposes served by the present invention in many àpplications. lmproved accuracy can be achieved by using an alternative embodiment to that illustrated in Figure 3. For example, peak estim~tes received from peak estim~tors 304 and 305 could be combined, passed to one peak hold c~",ponent and subsequently processed by threshold compare 308.
In an embodiment of the present invention for applications such as the broadcast STL, the signal received from peak hold 307 re~,resents the peak amplitude of frequencies below about S kHz. The peak amplitude estim~te for frequencies above about S kHz is received from peak hold 306 via threshold co~"pare 308. The peak amplitude of the full-bandwidth input signal at time n is ~cum~d to be the sum of these peak amplitude estim~t~s at time n, or P7~n) = PL(n) + PH(n) where PL(n) = held peak-amplitude ç~tim~tP for low frequencies at time n, and PH(n) = held peak-amplitude estim~te for high frequencies at time n.
For this particular application, it is ~sumed that only the high frequencies contain PLI, the~fo,t;, gain reduction should be applied only to these frequencies.
Whenever P7~n) el~ce~lc the reference level, gain select 310 establishes a gain factor g(n) to apply to the high frequencies in order to reduce the full-bandwidth peak amplitude to substantially the reference level. This gain factor may be obtained by solving the following two indep~ndent linear equations in two unknowns. These equations are eAI)~ssed in terms of peak-amplitude estim~tes which are norm~li7çd or scaled by the desired reference level such that the norm~li7e~ desired peak amplitude is equal to one; thus:
PL(n) + ph~n) = pT~n) pL(n) + g(n) pH(n) = 1.0 where pL(n) = norm~li7ed low frequency held peak-amplitude çstim~te at time n, pN(n) = norm~li7~d high frequency held peak-amplitude estim~te at time n, p7(n) = norm~li7~d full-bandwidth signal peak-amplitude estim~t~ at time n, and g(n) = gain factor required to optimally correct PLI at time n.
By solving these two equations simultaneously, the gain factor is seen to be Wo 94/19883 21 5 ~ 8 8 3 PCT/US94/01639 g(n) = I _ p7l~n)-l l-pL(n) pH(n) pH(n) (3) The gain table discu~ed above can be used to provide a gain factor which b~l~nces PLI
reduction against loss of ap~)arent loudness in the output signal. Conceptually, gain factor g(n) is the ratio between the high-frequency signal's "proper" level and its peak amplitude. This 5 ratio can be ~xpressed as 1/(1 +e) where e is the normalized deviation between the norm~li7ed high frequency signal peak amplitude p~(n) and the norm~li7ed amplitude required to correct PLI. Referring to equation (2), the index i into gain table T[i] is seen to be the actual deviation between the actual peak-amplitude estim~te x(n) and the actual reference level A,~,.
The actual deviation may be obtained by scaling the normalized deviation e with the actual 10 reference level A~". The.efore, gain select 310 may obtain a gain factor from the gain table entry TleA~,J.
Although the gain select function directly affects the ~,rol~l,ance of a particular embodiment, it should be understood that no particular gain select function is critical to the practice of the present invention. Other duplex gain select functions are possible and will be 15 ap~arent to those with oldinaly skill in the art.
e. Control Filter The dynamic characteristics of the control system must be established by b~l~ncing two conflicting goals. On the one hand, the control system should respond quickly to the onset of PLI and recover quickly after PLI has stopped so that the a~ale,-t loudness of the reproduced 20 signal is not affected more than nPces~ry. On the other hand, the conkol system should not respond and recover so quickly that it generates audible modulation distortion in the reproduced signal. The particular embodiment shown in Figure 3 from input path 302 up through gain select 310 allows the control system to respond and recover as quickly as possible. Control filter 312, however, restricts the rate of response to the generation of 25 audible artifacts.
Referring to Figure 1, gain element 114 responds to a gain control signal received from control system 122 to amplitude modulate the signal arriving along path 112. As a result of the arnplitude modul~tion, distortion in the form of ~ideb~n~c is generated.
Control filter 312 shown in Figure 3 is essentially a low-pass filter de~ign~ to limit 30 the frequency of the amplitude modulation and thereby limit the bandwidth of the sideb~nds.
The LPF bandwidth of control filter 312 should be chosen so that the sideband bandwidths are confined to within a psychoacoustic critical bandwidth of the spectral cG"~nents mocl~ ted wo 94/1g883 1 5 ~ 8 8 ~ PCT/US94/01639 by the variable gain element. In the preferred embodiment for the FM-broadcast STL
application (li~cllssed above, the function is implemented as a third order IIR filter with a cutoff frequency of 880 Hz, a stopband attenuation of -40 dB, and 0.09 dB of ripple in the low-pass band.
By ex~mining the effects of the modulation sidebands with respect to the psychoacoustic m~c1~ing thresholds of the modul~ted signal, the characteristics of the control filter may be specified so as to permit the fastest control system response possible consistent with psychoacoustic m~ ing of the modulation artifacts.
C. Sl~ lur~ of Alternate EmboJi~ s Many variations in embodiments of the present invention are possible. The following describes dirrerences between the alternate embodiments and the basic structure described above.
One embodiment of control system 122 is shown in Figure 6. Referring to Figure 6, 604 esli,n~s the peak ~mplitude of the input signal received from path 602 and passes the estim~te to peak hold 606. Peak hold 606 passes to threshold co~nl)a~e 608 a pulse-shaped signal which holds the estim~ted peak amplitude for a specific duration. Threshold colllpalc 608 colllpalcs the ~mplitude of the peak hold signal with a reference level and passes the results of the colllpdlison to gain select 610. Gain select 610 selects a gain factor in response to the signals received from threshold COlllp~UC 608 and passes the sele~ted gain factor to control filter 612. Control filter 612 ge.-e,dtes along path 614 a gain control signal in response to the s~ ted gain factor.
The functiQnc ~ro.llled by peak estim~t~r 604, peak hold 606 and threshold COIllpalc 608 are 5~5~ 1ly similar to the functions described above for the collc~nding el~ment~
shown in Figure 3 along the path carrying the first frequency subband con,ponent. Unlike the cou"te~ threshold colllpare 308, however, threshold col,lpa,c 608 receives an estim~te of the full-bandwidth input signal directly from one peak hold signal. Furthermore, threshold co,llparc 608 need not pass any peak hold signal to gain select 610, but can pass only the difference between the peak hold signal and the reference level.
Gain select 610 establishes a gain factor in response to the output of threshold co,llpalc 608. If the peak-amplitude estim~te is less than the reference level, a gain factor equal to one is select~. If the peak-amplitude e~l;",~te exceeds the reference level, however, a gain factor is select~ which reduces the amplitude of the frequency subband colllponent cont~ining PLI.
~_ 17 2154883 The gain select function may be implemented using equation (1) above, by selecting a gain factor equal to the ratio between the reference level and the peak-amplitude estim~e, or by using a gain table such as that shown in equation (2). Although the choice of gain select function may greatly affect system pe~ro~ aulce, no particular function is critical to the 5 practice of the present invention.
It should be appreciated that gain select 610 uses a simplex gain select function which generates a gain control signal in response to the full-bandwidth input signal. Other embodiments inco,~,d~ g duplex gain select functions such as those described above may generate a more optimal gain factor by responding to signal levels in both the full-bandwidth 10 and in the portion of the bandwidth in which PLI is to be limited. Although the implcm~nt~tion shown in Figure 6 may be suboptimal, it provides acceptable performance for a variety of applications including the broadcast STL application di~cucc~od above, and it ,~ui,es fewer proceccing resources than those required by the more optimum duplex embo-iiments.
Figure 7 illustrates an alternative but equivalent embodiment to that shown in Figure 1.
As shown in Figure 7, the splitter is imple--,æl-~ed with a low-pass filter and a subtractor, and the combiner is implement~ with an adder. LPF 708 receives the delayed full-bandwidth signal from path 706 and passes the low-frequency spectrum along path 710 to subtractor 711 and combiner 718. Subtractor 711 receives the delayed full-bandwidth signal from path 706, 20 subtracts the low-frequency col"ponents from the full-bandwidth co",~nents and passes the resl-lting signal having only high-frequency co"lponents along path 712 to gain elemt nt 714.
Gain elPm~nt 714 receives a gain control signal from control system 722 along path 724, applies a variable gain factor to the signal received from path 712, and passes the result to CGIllbiner 718. Colllbi~er 718 adds the signals received from paths 710 and 716 to generate 25 a full-bandwidth output signal along path 720.
Figure 8 illustrates an alternative embodiment to that shown in Figure 1 for imple...~ l;onc inco,~ldting a duplex gain select function. The functions pe,rol",ed by splitter 804 and delay elements 808 and 814 are interchanged as collll)~cd to that shown in Figure 1. As a result, control system 826 does not require a sc;pal~te splitter. The gain select 30 function can directly apply equation (3), ~liccucced above, to the first and second frequency subband co"")onents received from paths 812 and 806, res~e~;lively.
Figure 9 illustrates an alternative embodiment using a hybrid limiter structure similar to that shown in Figure 10c. Delay element 904 receives a full-bandwidth signal from path 902. Splitter 906 splits the delayed full-bandwidth signal into a first frequency subband WO 94/19883 21 S ~ 8 ~ 3 PCT/US94/01639 co-,-ponent, passed along path 910, and a second frequency subband component passed along path 908. Gain element 912 receives the first frequency subband co.n~nent from path 910, applies an adaptive gain factor to the first frequency subband co---pol-el1t and passes the result along path 914 to combiner 916. Combiner 916 combines the first and second frequency S subband co---ponents received from paths 914 and 908, respectively, and generates an output signal along path 918. Splitter 920 splits the fuli-bandwidth signal received from path 902 into a first frequency subband co---ponent, passed along path 922, and a second frequency subband co.-.~nent passed along path 928. Gain element 924 applies an adaptive gain factor to the first frequency subband co,.-ponent and passes the result to combiner 930. Combiner 930 10 combines the first and second frequency subband components received from paths 926 and 928, l~s~ ely, and generates a quasi-output signal along path 932. Control system 934 receives the quasi-output signal from path 932 and passes a gain control signal along path 936 to gain elements 912 and 924.
~ use output-controlled limiters are incpnsitive to precise control system 15 characteristics, embo~imentc of the present invention according to Figure 9 do not require accurate gain tables such as that .liccllcspd above.
The present invention may also be applied to multi-ch~nnPl applications. In one digital embo~limPnt incol~ld~ing a control system in accordal~ce with the structure shown in Figure 6, the maximum peak-amplitude estim~te across all channels is ~rtcented to the input 20 of a single peak hold 606. The rennlinder of the control system ~lrOl...s in a manner s,ll.sl.n~i~lly the same as that described above.
In an alternative embodiment which requires fewer plocPcc;ng resources, the largest m~gnitude sample for all ch~nn~plc is passed along path 602 to a single peak estim~tor 604.
Peak estimator 604 interpolates these largest m~nitude samples and passes a peak-amplitude 25 esli.n~te to peak hold 606.
It will be appar~;--t that the various functions in the control system structure as shown in Figure 6 may be shared across multiple ch~nnPlc in a similar manner as desired.
Many other control system embodiments are possible for multi-ch~nnel applications;
however, each must balance reduction in appalent loudness of each rh~nnçl, distortion of 30 relative ch~nnçl loudness, and modulation of frequency subband co..-ponents contdining PLI
by other frequency subband co~--ponents without PLI. For example, the maximum PLI-to-reference level ratio across all ch~nnels may be used to select a gain factor for each ch~nnel using the duplex gain select function described above. In this case, the duplex gain select wo 94/19883 215 4 8 8 3 PCT/US94/01639 function is applied using the peak-amplitude estim~tçs for each respective channel in combination with the common maximum PLI ratio.
The alternatives described above are given by way of example only, and illustrate that the present invention is applicable to a broad variety of structures and implementations.
Claims (15)
1. A signal processing system for processing an input audio signal limited to a first peak amplitude, said system comprising processing means (206, 210) for generating a processed audio signal in response to said input audio signal, wherein said processing means preserves some measure of spectral amplitude of said input audio signal but changes amplitude and/or phase in a portion of the total bandwidth of said processed audio signal causing the amplitude of said processed audio signal to exceed said first peak amplitude, control means (122; 722; 804, 826; 920, 924, 930, 9341 for generating an estimated peak amplitude of said processed audio signal and for establishing a gain factor in response to said estimated peak amplitude, and limiting means (108, 114, 118; 708, 711, 714, 718; 804, 818, 822; 906, 912, 916)responsive to said processed audio signal for generating an output audio signal limited to a second peak amplitude by applying said gain factor to said portion of the total bandwidth of said processed audio signal.
2. A signal processing system according to claim 1 wherein said processing meanscomprises a perceptual-coding system.
3. A signal processing system according to claim 1 or 2 wherein said processed audio signal comprises signal samples and wherein said control means comprises means (404, 406) for upsampling said signal samples.
4. A signal processing system according to any one of claims 1 through 3 whereinsaid control means generates said estimated peak amplitude in response to the full bandwidth of said processed audio signal.
5. A signal processing system according to claim 1 or 2 wherein said control means generates said estimated peak amplitude in response to a peak amplitude estimate of a sub-signal representing said processed audio signal within a frequency subband.
6. A signal processing system according to claim 5 wherein said peak amplitude estimate of said sub-signal is established by upsampling said sub-signal.
7. A signal processing system according to claim 1 or 2 wherein said control means comprises means (303-305) for generating said estimated peak amplitude in response to a first peak amplitude estimate of a first sub-signal representing said processed audio signal within a first frequency subband and in response to a second peak amplitude estimate of a second sub-signal representing said processed audio signal within a second frequency subband.
8. A signal processing system according to claim 7 wherein said first peak amplitude estimate is established by upsampling said first sub-signal.
9. A signal processing system according to any one of claims 1 through 7 whereinsaid control means further comprises peak hold means (306, 307; 606) for generating a peak-hold signal by holding said estimated peak amplitude, threshold compare means (308; 608) for comparing said peak-hold signal with a reference level and for generating a gain select signal in response thereto, andgain select means (310; 610) for establishing said gain factor in response to said gain select signal.
10. A signal processing system according to any one of claims 1 through 9 wherein said control means further comprises means (312; 612) for controlling the rate of change of said gain factor so as to control the audibility of resulting artifacts.
11. A signal processing system for processing an input audio signal limited to a first peak amplitude limit, said system comprising a split-band encoder (206) having an output and an input coupled to said input audio signal, a split-band decoder (210) having an output and an input coupled to the output of said split-band encoder, a peak-amplitude estimator (604, 606, 608) having an output and an input coupled to the output of said split-band decoder, a gain control (610, 612) having an output and an input coupled to the output of said peak amplitude estimator, a first filter (108) having an output and an input coupled to the output of said split-band decoder, one or more second filters (108) each having an output and an input coupled to the output of said split-band decoder, a gain element (114) with variable gain having an output, a signal input coupled to the output of said first filter, and a gain input coupled to the output of said gain control, wherein said variable gain is responsive to said gain input, a combiner (118) having a system output, an input coupled to the output of said gain element, and a respective input coupled to the output of each of said one or more second filters, and means (104) for delaying, relative to the output of said split-band decoder, signals generated by said first filter and said one or more second filters.
12. A signal processing system according to claim 11 wherein said peak-amplitudeestimator comprises a third filter (303) having a third filter output and an input coupled to the input of said peak-amplitude estimator, a fourth filter (303) having a fourth filter output and an input coupled to the input of said peak-amplitude estimator, a first peak-amplitude estimator (305, 307) having an output and an input coupled to said third filter output, a second peak-amplitude estimator (304, 306) having an output and an input coupled to said fourth filter output, and a combiner (308) having an output coupled to the output of said peak-amplitude estimator, an input coupled to the output of said first peak-amplitude estimator, and an input coupled to the output of said second peak-amplitude estimator.
13. A signal processing system for processing an input audio signal limited to a first peak amplitude limit, said system comprising a split-band encoder (206) having an output and an input coupled to said input audio signal, a split-band decoder (210) having an output and an input coupled to the output of said split-band encoder, a first filter (804) having an output and an input coupled to the output of said split-band decoder, one or more second filters (804) each having an output and an input coupled to the output of said split-band decoder, a first peak-amplitude estimator (305, 307, 308) having an output and an input coupled to the output of said first filter, a second peak-amplitude estimator (304, 306, 308) having an output and an input coupled to the output of each of said one or more second filters, a gain control (310, 312) having an output and an input coupled to the output of said first peak amplitude estimator and to the output of said second peak-amplitude estimator, a first delay (814) having an output and an input coupled to the output of said first filter, one or more second delays (808) each having an output and an input coupled to the output of a respective one of said one or more second filters, a gain element (818) with variable gain having an output, a signal input coupled to the output of said first delay, and a gain input coupled to the output of said gain control, wherein said variable gain is responsive to said gain input, and a combiner (822) having a system output, an input coupled to the output of said gain element, and a respective input coupled to each output of said one or more second delays.
14. A signal processing system according to any one of claims 11 through 13 wherein said split-band decoder (210) generates an output signal comprising samples, and wherein said peak-amplitude estimator further comprises one or more upsampling filters (404, 406).
15. A signal processing system according to any one of claims 11 through 14 wherein said gain control (310; 312; 610, 612) further comprises a filter (312; 612) coupling the output of said gain control to the gain input of said gain element (114; 818).
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US08/017,535 US5579404A (en) | 1993-02-16 | 1993-02-16 | Digital audio limiter |
| US08/017,535 | 1993-02-16 | ||
| PCT/US1994/001639 WO1994019883A1 (en) | 1993-02-16 | 1994-02-15 | Digital audio limiter |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CA2154883A1 true CA2154883A1 (en) | 1994-09-01 |
Family
ID=21783140
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA002154883A Abandoned CA2154883A1 (en) | 1993-02-16 | 1994-02-15 | Digital audio limiter |
Country Status (7)
| Country | Link |
|---|---|
| US (1) | US5579404A (en) |
| EP (1) | EP0685130B1 (en) |
| JP (1) | JPH08507186A (en) |
| AU (1) | AU686602B2 (en) |
| CA (1) | CA2154883A1 (en) |
| DE (1) | DE69400819T2 (en) |
| WO (1) | WO1994019883A1 (en) |
Families Citing this family (82)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7424731B1 (en) | 1994-10-12 | 2008-09-09 | Touchtunes Music Corporation | Home digital audiovisual information recording and playback system |
| US7188352B2 (en) | 1995-07-11 | 2007-03-06 | Touchtunes Music Corporation | Intelligent digital audiovisual playback system |
| US8661477B2 (en) | 1994-10-12 | 2014-02-25 | Touchtunes Music Corporation | System for distributing and selecting audio and video information and method implemented by said system |
| ATE188793T1 (en) | 1994-10-12 | 2000-01-15 | Touchtunes Music Corp | INTELLIGENT SYSTEM FOR NUMERICAL AUDIOVISUAL REPRODUCTION |
| GB9500285D0 (en) * | 1995-01-07 | 1995-03-01 | Central Research Lab Ltd | A method of labelling an audio signal |
| US5666430A (en) * | 1995-01-09 | 1997-09-09 | Matsushita Electric Corporation Of America | Method and apparatus for leveling audio output |
| US5812969A (en) * | 1995-04-06 | 1998-09-22 | Adaptec, Inc. | Process for balancing the loudness of digitally sampled audio waveforms |
| KR100463462B1 (en) | 1995-10-24 | 2005-05-24 | 코닌클리케 필립스 일렉트로닉스 엔.브이. | Repeated decoding and encoding in subband encoder/decoders |
| KR100196425B1 (en) * | 1996-05-16 | 1999-06-15 | 윤종용 | Method for controlling sound of multimedia display monitor |
| JPH1083623A (en) * | 1996-09-10 | 1998-03-31 | Sony Corp | Signal recording method, signal recorder, recording medium and signal processing method |
| FR2753868A1 (en) | 1996-09-25 | 1998-03-27 | Technical Maintenance Corp | METHOD FOR SELECTING A RECORDING ON AN AUDIOVISUAL DIGITAL REPRODUCTION SYSTEM AND SYSTEM FOR IMPLEMENTING THE METHOD |
| FR2769165B1 (en) | 1997-09-26 | 2002-11-29 | Technical Maintenance Corp | WIRELESS SYSTEM WITH DIGITAL TRANSMISSION FOR SPEAKERS |
| DE19824233B4 (en) * | 1998-05-29 | 2005-10-06 | Telefonaktiebolaget Lm Ericsson (Publ) | amplitude limiting |
| FR2781582B1 (en) | 1998-07-21 | 2001-01-12 | Technical Maintenance Corp | SYSTEM FOR DOWNLOADING OBJECTS OR FILES FOR SOFTWARE UPDATE |
| US8028318B2 (en) | 1999-07-21 | 2011-09-27 | Touchtunes Music Corporation | Remote control unit for activating and deactivating means for payment and for displaying payment status |
| FR2781580B1 (en) | 1998-07-22 | 2000-09-22 | Technical Maintenance Corp | SOUND CONTROL CIRCUIT FOR INTELLIGENT DIGITAL AUDIOVISUAL REPRODUCTION SYSTEM |
| FR2781591B1 (en) | 1998-07-22 | 2000-09-22 | Technical Maintenance Corp | AUDIOVISUAL REPRODUCTION SYSTEM |
| DE69813912T2 (en) * | 1998-10-26 | 2004-05-06 | Stmicroelectronics Asia Pacific Pte Ltd. | DIGITAL AUDIO ENCODER WITH VARIOUS ACCURACIES |
| US6757396B1 (en) * | 1998-11-16 | 2004-06-29 | Texas Instruments Incorporated | Digital audio dynamic range compressor and method |
| US6337999B1 (en) * | 1998-12-18 | 2002-01-08 | Orban, Inc. | Oversampled differential clipper |
| US8726330B2 (en) | 1999-02-22 | 2014-05-13 | Touchtunes Music Corporation | Intelligent digital audiovisual playback system |
| DE60040598D1 (en) | 1999-04-16 | 2008-12-04 | Orban Inc | Halbcosinus-interpolation |
| FR2796482B1 (en) | 1999-07-16 | 2002-09-06 | Touchtunes Music Corp | REMOTE MANAGEMENT SYSTEM FOR AT LEAST ONE AUDIOVISUAL INFORMATION REPRODUCING DEVICE |
| FR2805377B1 (en) | 2000-02-23 | 2003-09-12 | Touchtunes Music Corp | EARLY ORDERING PROCESS FOR A SELECTION, DIGITAL SYSTEM AND JUKE-BOX FOR IMPLEMENTING THE METHOD |
| FR2805072B1 (en) | 2000-02-16 | 2002-04-05 | Touchtunes Music Corp | METHOD FOR ADJUSTING THE SOUND VOLUME OF A DIGITAL SOUND RECORDING |
| FR2805060B1 (en) | 2000-02-16 | 2005-04-08 | Touchtunes Music Corp | METHOD FOR RECEIVING FILES DURING DOWNLOAD |
| DE10012003A1 (en) * | 2000-03-11 | 2001-09-13 | Philips Corp Intellectual Pty | Transmitter and method for generating a transmission signal |
| FR2808906B1 (en) | 2000-05-10 | 2005-02-11 | Touchtunes Music Corp | DEVICE AND METHOD FOR REMOTELY MANAGING A NETWORK OF AUDIOVISUAL INFORMATION REPRODUCTION SYSTEMS |
| FR2811175B1 (en) | 2000-06-29 | 2002-12-27 | Touchtunes Music Corp | AUDIOVISUAL INFORMATION DISTRIBUTION METHOD AND AUDIOVISUAL INFORMATION DISTRIBUTION SYSTEM |
| FR2811114B1 (en) | 2000-06-29 | 2002-12-27 | Touchtunes Music Corp | DEVICE AND METHOD FOR COMMUNICATION BETWEEN A SYSTEM FOR REPRODUCING AUDIOVISUAL INFORMATION AND AN ELECTRONIC ENTERTAINMENT MACHINE |
| FR2814085B1 (en) | 2000-09-15 | 2005-02-11 | Touchtunes Music Corp | ENTERTAINMENT METHOD BASED ON MULTIPLE CHOICE COMPETITION GAMES |
| US20020075965A1 (en) * | 2000-12-20 | 2002-06-20 | Octiv, Inc. | Digital signal processing techniques for improving audio clarity and intelligibility |
| US7242784B2 (en) * | 2001-09-04 | 2007-07-10 | Motorola Inc. | Dynamic gain control of audio in a communication device |
| US7376159B1 (en) | 2002-01-03 | 2008-05-20 | The Directv Group, Inc. | Exploitation of null packets in packetized digital television systems |
| EP1472786A2 (en) * | 2002-01-24 | 2004-11-03 | Koninklijke Philips Electronics N.V. | A method for decreasing the dynamic range of a signal and electronic circuit |
| US7286473B1 (en) | 2002-07-10 | 2007-10-23 | The Directv Group, Inc. | Null packet replacement with bi-level scheduling |
| US8151304B2 (en) | 2002-09-16 | 2012-04-03 | Touchtunes Music Corporation | Digital downloading jukebox system with user-tailored music management, communications, and other tools |
| US7822687B2 (en) | 2002-09-16 | 2010-10-26 | Francois Brillon | Jukebox with customizable avatar |
| US9646339B2 (en) | 2002-09-16 | 2017-05-09 | Touchtunes Music Corporation | Digital downloading jukebox system with central and local music servers |
| US10373420B2 (en) | 2002-09-16 | 2019-08-06 | Touchtunes Music Corporation | Digital downloading jukebox with enhanced communication features |
| US8584175B2 (en) | 2002-09-16 | 2013-11-12 | Touchtunes Music Corporation | Digital downloading jukebox system with user-tailored music management, communications, and other tools |
| US11029823B2 (en) | 2002-09-16 | 2021-06-08 | Touchtunes Music Corporation | Jukebox with customizable avatar |
| US8103589B2 (en) | 2002-09-16 | 2012-01-24 | Touchtunes Music Corporation | Digital downloading jukebox system with central and local music servers |
| US8332895B2 (en) | 2002-09-16 | 2012-12-11 | Touchtunes Music Corporation | Digital downloading jukebox system with user-tailored music management, communications, and other tools |
| US6760452B2 (en) * | 2002-10-24 | 2004-07-06 | Visteon Global Technologies, Inc. | Multi-channel audio signal limiter with shared clip detection |
| US7647221B2 (en) * | 2003-04-30 | 2010-01-12 | The Directv Group, Inc. | Audio level control for compressed audio |
| US7912226B1 (en) | 2003-09-12 | 2011-03-22 | The Directv Group, Inc. | Automatic measurement of audio presence and level by direct processing of an MPEG data stream |
| US7202731B2 (en) * | 2005-06-17 | 2007-04-10 | Visteon Global Technologies, Inc. | Variable distortion limiter using clip detect predictor |
| US8166416B2 (en) * | 2005-08-17 | 2012-04-24 | Cyber Group Usa, Inc. | Play menu and group auto organizer system and method for a multimedia player |
| WO2007098258A1 (en) * | 2006-02-24 | 2007-08-30 | Neural Audio Corporation | Audio codec conditioning system and method |
| US7983425B2 (en) * | 2006-06-13 | 2011-07-19 | Phonak Ag | Method and system for acoustic shock detection and application of said method in hearing devices |
| JP4972742B2 (en) * | 2006-10-17 | 2012-07-11 | 国立大学法人九州工業大学 | High-frequency signal interpolation method and high-frequency signal interpolation device |
| US9171419B2 (en) | 2007-01-17 | 2015-10-27 | Touchtunes Music Corporation | Coin operated entertainment system |
| US9330529B2 (en) | 2007-01-17 | 2016-05-03 | Touchtunes Music Corporation | Game terminal configured for interaction with jukebox device systems including same, and/or associated methods |
| US9953481B2 (en) | 2007-03-26 | 2018-04-24 | Touchtunes Music Corporation | Jukebox with associated video server |
| KR101355376B1 (en) * | 2007-04-30 | 2014-01-23 | 삼성전자주식회사 | Method and apparatus for encoding and decoding high frequency band |
| US8332887B2 (en) | 2008-01-10 | 2012-12-11 | Touchtunes Music Corporation | System and/or methods for distributing advertisements from a central advertisement network to a peripheral device via a local advertisement server |
| US10290006B2 (en) | 2008-08-15 | 2019-05-14 | Touchtunes Music Corporation | Digital signage and gaming services to comply with federal and state alcohol and beverage laws and regulations |
| WO2009086174A1 (en) | 2007-12-21 | 2009-07-09 | Srs Labs, Inc. | System for adjusting perceived loudness of audio signals |
| KR20100134623A (en) * | 2008-03-04 | 2010-12-23 | 엘지전자 주식회사 | Method and apparatus for processing an audio signal |
| WO2009110087A1 (en) * | 2008-03-07 | 2009-09-11 | ティーオーエー株式会社 | Signal processing device |
| WO2010005569A1 (en) | 2008-07-09 | 2010-01-14 | Touchtunes Music Corporation | Digital downloading jukebox with revenue-enhancing features |
| US9292166B2 (en) | 2009-03-18 | 2016-03-22 | Touchtunes Music Corporation | Digital jukebox device with improved karaoke-related user interfaces, and associated methods |
| US10719149B2 (en) | 2009-03-18 | 2020-07-21 | Touchtunes Music Corporation | Digital jukebox device with improved user interfaces, and associated methods |
| KR101748448B1 (en) | 2009-03-18 | 2017-06-16 | 터치튠즈 뮤직 코포레이션 | Entertainment server and associated social networking services |
| US10564804B2 (en) | 2009-03-18 | 2020-02-18 | Touchtunes Music Corporation | Digital jukebox device with improved user interfaces, and associated methods |
| US8538042B2 (en) | 2009-08-11 | 2013-09-17 | Dts Llc | System for increasing perceived loudness of speakers |
| US20120278087A1 (en) * | 2009-10-07 | 2012-11-01 | Nec Corporation | Multiband compressor and method of adjusting the same |
| KR101633709B1 (en) * | 2010-01-12 | 2016-06-27 | 삼성전자주식회사 | Method and apparatus for removing acoustic incident |
| CA2881456A1 (en) | 2010-01-26 | 2011-08-04 | Touchtunes Music Corporation | Digital jukebox device with improved user interfaces, and associated methods |
| US9620131B2 (en) | 2011-04-08 | 2017-04-11 | Evertz Microsystems Ltd. | Systems and methods for adjusting audio levels in a plurality of audio signals |
| US9729120B1 (en) | 2011-07-13 | 2017-08-08 | The Directv Group, Inc. | System and method to monitor audio loudness and provide audio automatic gain control |
| GB2522772B (en) | 2011-09-18 | 2016-01-13 | Touchtunes Music Corp | Digital jukebox device with karaoke and/or photo booth features, and associated methods |
| US11151224B2 (en) | 2012-01-09 | 2021-10-19 | Touchtunes Music Corporation | Systems and/or methods for monitoring audio inputs to jukebox devices |
| US9173025B2 (en) | 2012-02-08 | 2015-10-27 | Dolby Laboratories Licensing Corporation | Combined suppression of noise, echo, and out-of-location signals |
| US8712076B2 (en) | 2012-02-08 | 2014-04-29 | Dolby Laboratories Licensing Corporation | Post-processing including median filtering of noise suppression gains |
| US9312829B2 (en) | 2012-04-12 | 2016-04-12 | Dts Llc | System for adjusting loudness of audio signals in real time |
| US9735746B2 (en) * | 2012-08-01 | 2017-08-15 | Harman Becker Automotive Systems Gmbh | Automatic loudness control |
| EP2693635A1 (en) * | 2012-08-01 | 2014-02-05 | Harman Becker Automotive Systems GmbH | Automatic loudness control |
| JP6056356B2 (en) * | 2012-10-10 | 2017-01-11 | ティアック株式会社 | Recording device |
| WO2015070070A1 (en) | 2013-11-07 | 2015-05-14 | Touchtunes Music Corporation | Techniques for generating electronic menu graphical user interface layouts for use in connection with electronic devices |
| EP3123293A4 (en) | 2014-03-25 | 2017-09-27 | Touchtunes Music Corporation | Digital jukebox device with improved user interfaces, and associated methods |
Family Cites Families (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA1028627A (en) * | 1975-08-08 | 1978-03-28 | Robert A. Orban | Method and system for controlling peak signal levels in a bandlimited recording or transmissions system employing high-frequency pre-emphasis |
| JPS56122243A (en) * | 1980-02-29 | 1981-09-25 | Victor Co Of Japan Ltd | Noise reduction system |
| US4337445A (en) * | 1981-01-21 | 1982-06-29 | Sony Corporation | Compander circuit which produces variable pre-emphasis and de-emphasis |
| US4406923A (en) * | 1981-10-28 | 1983-09-27 | Cbs Inc. | Automatic loudness controller |
| US4604755A (en) * | 1984-06-01 | 1986-08-05 | International Business Machines Corp. | Feed forward dual channel automatic level control for dual tone multi-frequency receivers |
| US4701953A (en) * | 1984-07-24 | 1987-10-20 | The Regents Of The University Of California | Signal compression system |
| US4882761A (en) * | 1988-02-23 | 1989-11-21 | Resound Corporation | Low voltage programmable compressor |
| KR910007982Y1 (en) * | 1989-09-19 | 1991-10-10 | 삼성전자 주식회사 | Middle-low level sound control circuit |
| US5040217A (en) * | 1989-10-18 | 1991-08-13 | At&T Bell Laboratories | Perceptual coding of audio signals |
| US5020098A (en) * | 1989-11-03 | 1991-05-28 | At&T Bell Laboratories | Telephone conferencing arrangement |
| US5050217A (en) * | 1990-02-16 | 1991-09-17 | Akg Acoustics, Inc. | Dynamic noise reduction and spectral restoration system |
| US5371803A (en) * | 1990-08-31 | 1994-12-06 | Bellsouth Corporation | Tone reduction circuit for headsets |
| US5278912A (en) * | 1991-06-28 | 1994-01-11 | Resound Corporation | Multiband programmable compression system |
-
1993
- 1993-02-16 US US08/017,535 patent/US5579404A/en not_active Expired - Lifetime
-
1994
- 1994-02-15 DE DE69400819T patent/DE69400819T2/en not_active Expired - Fee Related
- 1994-02-15 EP EP94909637A patent/EP0685130B1/en not_active Expired - Lifetime
- 1994-02-15 JP JP6519089A patent/JPH08507186A/en active Pending
- 1994-02-15 CA CA002154883A patent/CA2154883A1/en not_active Abandoned
- 1994-02-15 AU AU62406/94A patent/AU686602B2/en not_active Ceased
- 1994-02-15 WO PCT/US1994/001639 patent/WO1994019883A1/en active IP Right Grant
Also Published As
| Publication number | Publication date |
|---|---|
| AU686602B2 (en) | 1998-02-12 |
| US5579404A (en) | 1996-11-26 |
| DE69400819D1 (en) | 1996-12-05 |
| EP0685130B1 (en) | 1996-10-30 |
| AU6240694A (en) | 1994-09-14 |
| JPH08507186A (en) | 1996-07-30 |
| EP0685130A1 (en) | 1995-12-06 |
| WO1994019883A1 (en) | 1994-09-01 |
| DE69400819T2 (en) | 1997-04-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CA2154883A1 (en) | Digital audio limiter | |
| CN100362777C (en) | BTSC encoder | |
| CA2294140C (en) | A transmission apparatus capable of attenuating a peak power level prior to a signal output | |
| US7715567B2 (en) | Noise reduction in a stereo receiver | |
| EP0958699A2 (en) | Method and apparatus for recognising video sequences | |
| HUT72945A (en) | A method and apparatus for mitigating distortion effects in the dermination of signal usability | |
| KR20040035749A (en) | Bandwidth extension of a sound signal | |
| US5913152A (en) | FM composite signal processor | |
| EP1289155A1 (en) | Noise reduction in an FM stereo receiver | |
| US5815532A (en) | Method and apparatus for peak-to-average ratio control in an amplitude modulation paging transmitter | |
| US6618486B2 (en) | Controller for FM 412 multiplex power regulation | |
| FI79428C (en) | NAET FOER DAEMPNING AV MELLANFREKVENSMODULATIONSFENOMEN HOS KOMPRESSORER, EXPANDRAR OCH BRUSDAEMPNINGSSYSTEM. | |
| US5953067A (en) | Multichannel television sound stereo and surround sound encoder | |
| WO1998035495A9 (en) | Multichannel television sound stereo and surround sound encoder | |
| DK169693B1 (en) | Method and circuit arrangement for spectral correction and post-correction | |
| GB2385420A (en) | Measuring the perceived loudness of an audio signal | |
| KR100513543B1 (en) | BTSC encoder | |
| US7362872B2 (en) | Method for noise reduction of a FM signal | |
| AU775461B2 (en) | BTSC encoder | |
| JPH0520006B2 (en) | ||
| EP1367746A1 (en) | Compression of audio signals in a FM transmitter | |
| AU2007254652A1 (en) | BTSC encoder |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| FZDE | Discontinued | ||
| FZDE | Discontinued |
Effective date: 20020215 |