GB2229343A - Suppression of undesired spectra in digital dynamic expansion - Google Patents

Description

SUPPRESSION OF UNDESIRED SPECTRA IN DIGITAL DYNAMIC EXPANSION The present

invention relates to a method for suppression or attenuation of undesired spectra in dynamic expansion of, in particular, digitalised compressed signals and to signal processing means by which such method may be carried out.

In DE- P 38 15 079 there is described a digital dynamic expander used, after an analog-to-digital conversion, to expand compressed analog signals recorded on a storage medium. A prerequisite for this method is a digital expander which satisfactorily statically and dynamically reverses the compressor function utilised in the preceding analog compression and which also replicates a suitable analog expander in digital technique.

An expander can restore an original signal, i.e. to the-form it had before the compression, only apart from certain residual errors. Such cases manifest themselves at the output of an analog expander by the appearance of predominantly odd-numbered harmonic spectra] components which, however, for example in.the case of sound signals, are not aurally perceived as particularly disturbing. The tone colour of the sound signals is changed only very slightly by these harmonic spectra.

Thereagainst, non-harmonic spectra, which can be very disturbing, arise in digital expansion.

There is therefore a need to suppress or attenuate undesired spectra in digital dynamic expansion.

According to one aspect of the present invention there is provided a method for suppression or attenuation of undesired spectra in dynamic expansion of digitalised compressed signals, the method comprising_the step of using a higher scanning frequency for the digital dynamic expansion than for subsequent digital further processing of the expanded signals.

According to another aspect of the present invention there is provided signal processing means comprising an arrangement for suppression or attenuation of undesired spectra in a digital dynamic expander, the arrangement comprising first means to increase the scanning frequency of unexpanded signals relative to a given normal scanning frequency by which the signals are processed after expansion in the expander, and second means to subsequently reduce the scanning frequency to the normal scanning frequency.

Examples of the method and embodiments of signal processing means of the present invention will now be more particularly described with reference to the accompanying drawings, in which:

Fig. 1 is a block schematic diagram of a digital dynamic expander; Fig. 2 is a spectral diagram; Fig. 3 is a block circuit diagram of first signal processing means embodying the invention; Fig. 4 is a block circuit diagram of second signal processing means embodying the invention; Fig. 5 is a block circuit diagram of third signal processing means embodying the invention; and Fig. 6 is a block circuit diagram of an expander measurement path in fourth signal processing means embodying the invention.

Referring now to the drawings, there is shown in Fig. 1 a digital dynamic expander comprising branches 1,..., 1, 1+1,..., L, the inputs of which are denoted by El,..., E], El+l,..., EL, which branch off from an input terminal Eo. The branches, of which only the branch 1 is illustrated in detail, all have basically the same structure, but differ in general by parameters which, in particular, determine the respective dynamic behaviour. In addition, filters FS and FM, which are provided at the inputs, for the splitting of partial frequency bands over the different branches are different.

Either or each of the filters FS and FM can consist of cascaded or parallelly connected part filters. The filter can be a digital recursive filter in canonic structure (cf H. G6ckler: Einstellbare Digitalfilter fUr die Tontechnik, ntz archive, volume 7 (1985), issue 3, pages 47 to 57, Figure 2) or in phase space structure (cf German (Fed.Rep.) patent applications 35 22 411,_35 22 412, 35 22 413 and 34 39 977).

Each branch divides at a junction Ab into a signal path S and a measurement path M. The branch signal separation is initially undertaken in the signal path S by the filter FS. The signal to be expanded is weighted (amplified, attenuated and so forth) by a signal path multiplier SM operating as a setting member, and in particular with a setting magnitude SG which is fed from the measurement path M into the second input of the signal path multiplier SM and which is ultimately derived from the signal at the input El. A further multiplication by a constant ks takes place in an equalising multiplier Ms for equalising purposes.

Finally, all branch signals, thus all signals at the expander out- puts Al,..., A], Al+l,..., AL are added by means of a digital adder Ad for formation of the digitally expanded total signal A at an output Ao.

A further band limitation in the filter FM takes place downstream of the junction Ab in the measurement path M. The output signal thereof feeds one of the inputs of a measurement path multiplier MM, which is part of a regulating circuit R. For equalising purposes, the output signal of the multiplier MM is weighted by a constant km in an equalising multiplier Mm. A magnitude formation operation (suppression of sign) takes place in a magnitude former B and the addition of a correction constant thereafter in an adder A]. An amplitude limiter SL, which is followed by a logarithmator LM, is also provided within the regulating circuit R. By virtue of the transition from the linear to the logarithmic value representation (]in/log), the expander characteristic is linear in the logarithmic range. The first operation in the logarithmic value range is a level-dependent amplification in a non-linear transmission member NU. Thereafter, an adder A2 is provided for the addition of a magnitude Io, and finally an integrator which has an amplitude limiter Sz in an integrating regulating path and a delay member T' in a feedback Rf to an adder A3. The output signal of the integrator is weighted by -a in a feedback branch of the regulating circuit R lead- ing back to the multiplier MM. A multiplier serves as a weighting means M1 in the feedback branch. The multiplier M1 is followed by a delogarithmator DG, by which the signals are again transformed back into the linear value range (log/]in conversion), thus an operation inverse to the linear-to-logarithmic conversion in the logarithmator LM. The output signals of the delogarithmator DG serve as setting magnitude within the regulating circuit R in the measurement path and are thus conducted to the multiplier MM.

The setting magnitude SG for the signal path S is derived from the output signal of the integrator and thereby of the regulating circuit R thorugh weighting by a(n-m)/m in a weighting means M2, namely a _ multiplier, with subsequent delogarithmation in a delogarithmator D2 in correspondence with that in the feedback branch of the regulating J l; circuit R. The parameters n and m determine the static behaviour, thus the slope n/m of the amplification characteristic, of the digital expander. The dynamic behaviour of the digital expander can be fixed by the choice of the parameter a.

The expander operates in such a manner that, for a constant input level at the input El, a signal which remains constant from scanning value to scanning value arises at the output of the regulating circuit R, thus the setting magnitude SG remains constant. However, these signals are dependent on level, so that higher signal levels in the signal path S are raised more strongly by the signal path multiplier SM than lower signal levels (or lower signals levels are lowered more strongly than higher ones), which is after all the purpose of an expander. In the case of upward or downward steps in level at the input E], certain modes of transient behaviour must appear at the output of the regulating path R in order that the desired original signal, as it was before the compression, arises at the signal path junction which is formed by the signal path multiplier SM. In that case it is necessary to take into account the transient behaviour of the compressor which produced the input signal now present and converted from analog to digital at the input Eo. This transient behaviour of the compressor usually differs in dependence on whether an upward or downward level step has taken place. In the case of an upward step, the transition time is very short (in the range of milliseconds), but relatively long in the case of a downward step (in the range of seconds). In order that the expander shown in Fig. 1 exhibits dynamic behaviour match-ed to the behaviour of the associated compressor, the non-linear transmission member NU, which effects a level-dependent amplification in the logarithmic weighting range, is provided in the regulating path. This, together with the downstream integrator, determines the dynamic behaviour of the expander during upward steps of the level. The constant Io, which is conducted to the adder A2, principally takes care of the dynamic behaviour during downward steps, thus of the socalled decay.

A form of digital dynamic expander, for example of the kind described in German (Fed. Rep.) patent application P 38 15 079, was described in the preceding simply by way of example. The measures, explained in the following, for suppression or attenuation of disturbing spectra can be applied to this expander or to any form of expander modified by comparison therewith.

During the signal processing in such a digital dynamic expander, nonlinear operations are also performed such as by the magnitude form- er B and by the non-linear amplifier (transmission member) NU in the described expander of Fig. 1. Such non-linear operations give rise to overtones which have a non-harmonic relationship to the useful signal spectrum and therefore have a disturbing effect. The particular interference spectra that arise from the digital dynamic expansion are explained by reference to the spectrum diagram S of Fig. 2. In this diagram, fA denotes the scanning frequency by which the signals are further processed after the digital dynamic expansion and this frequency fA i.s designated in the following as normal scanning frequency.

The useful spectrum, i.e. the spectrum of the signal to be expanded, is designated by fl. If, for example, a signal at the frequency fl is present at the input E] of the expander, then the harmonic frequencies 2kfl (k = 1, 2,...) result from this frequency fl in the measurement I :k i path M mainly due to the magnitude formation at the former B. These spectra in the output signal SG of the measurement path M are mixed in the multiplier SM with the still undisturbed input signal of the frequency fl on the signal path. The spectra f k = W 1 + f 1 = (2k +. 1) fl thereby result. These spectra represent only a small disturbance in the case of analog dynamic expansion, since they are basically harmonic in relation to the useful spectrum fl. However, if the scanning frequency within the expander in the case of a digital dynamic expansion has a non- harmonic relation to the frequency fl, non-harmonic image frequencies (fA - 3fl, 2fA - 5fl,...) arise due to the periodic nature of the spectra of scanned signals, i.e. the predominantly odd-numbered harmonics (2k + 1) fl of the input signal frequency fl are imaged at the (half and whole) scanning frequency or multiple thereof. Due to such imagings, only the harmonic spectra (2k + 1) fl above half the scanning frequency are converted into non-harmonic spectra.

It is thus desirable to provide suppression or the greatest possible attenuation of these disturbing non-harmonic spectra, which, in the following embodiments of the invention, is achieved by using a scanning frequency fAR, which is higher than the normal scanning frequency fA, for the digital dynamic expansion. The following aspects are decisive for fixing of the ratio fAl/fA: 1. For practical reasons, fA'/fA will generally (but not necessarily) be a whole number greater than 1. The ratio fAA/fA is selected to be so great that the non-harmonic interfering spectra imaged at the integral multiple of half (fA'/2) and the whole (fA') scanning frequency have such small amplitude values thattheir disturbance is negligible or at least tolerable. The prerequisite for this is that the amplitudes of the harmonic overtones decrease generally with the ordinal number.

Fig. 3 shows an embodiment in which an analog compressed signal is converted into a digital signal by an analog-to-digital converter A/D at a scanning frequency fA. higher than the normal scanning frequency fA. This digital signal is then expanded digitally at the increased scanning frequency fAl. A circuit arrangement, such as shown in Fig. 1, can be used for the digital dynamic expander EXP. At the output Ao of the expander EXP, the expanded signal with the increased scanning frequency fAl is fed to a decimation filter DF. The filter DF effects a lowering of the scanning frequency fAl to the normal scanning frequency fA at which subsequent signal processing takes place.

In the embodiment shown in Fig. 4, the compressed analog signal is converted in an analogto-digital converter A/D at a scanning frequency fA'., which is smaller or greater than,or equal to, the normal scanning frequency fA. A downstream interpolation filter IF increases the scanning frequency fA'' to fAi at which the expansion of the signal in the digital dynamic expander EXP is then performed. The output signal of the expander EXP is lowered in its scanning frequency in a decimation filter DF from fAl to fA.

Interpolation filters and decimation filters are described in R. E. Crochiere, L. R. Rabiner: Multi-rate Digital signal processing (Prentice Hall Inc. Englewood Cliffs, N.J., 1983).

It is advantageous, in the case of a digital dynamic expander. of the kind illustrated in Fig. 1, to undertake the increase in the scanning frequency from fA'' to fAl (fAl greater than fA) only after R 0 the input filter FS, because a lower scanning frequency of the signal to be processed by the filter imposes a smaller load on the filter. For that reason, the interpolation.filter IF is connected downstream of the input filter FS, as shown in Fig. 5. The lowering of the scanning frequency of the expander output signal from fAl to fA in the decimation filter DF expediently takes place at the output of the adder Ad, which adds the output signals, allotted to different frequency bands, of the parallelly connected expander units. This arrangement avoids the need for a respective decimation filter at the outputs Al, A], AL + 1... AL of the individual expander units.

A further improvement in respect of suppression of interference spectra may be achieved if the scannf ng- frequency is increased especially for the regulating circuit R in the measurement path M - compared with the scanning frequency fA' in the signal path S - to fAll' fAl. AS shown in Fig. 6, an interpolation filter IF1, which increases the scanning frequency fA' to fA''', is for this purpose connected downstream of the filter FM in the measurement path M. After the regulating circuit R, lowering of the high scanning frequency fAll' to the lower scanning frequency fAl otherwise used in the expander is effected by a decimation filter DFl.

Claims (1)

1. CLAIMS 1. A method for suppression or attenuation of undesired spectra in

   dynamic expansion of digitalised compressed signals, the method comprising the step of using a higher scanning frequency for the digital dynamic expansion than for subsequent digital further process ing of the expanded signals.

   2. A method as claimed in claim 1 and substantial] y as hereinbefore described with reference to any one of Figs. 3 to 6 of the accompanying drawings.

   3. Signal processing means comprising an arrangement for suppression or attenuation of undesired spectra in a digital dynamic expander, the arrangement comprising first means to increase the scanning frequency of unexpanded signals relative to a given normal scanning frequency by which the signals are processed after expansion in the expander, and second means to subsequently reduce the scanning frequency to the normal scanning frequency.

   4. Signal processing means as claimed in claim 3, wherein the expander has a signal path and a measurement path which branches off from the signal path and includes a regulating circuit having an output coupled to the signal path downstream of a frequency filter therein, the filter having a pass band matched to the frequency band in which dynamic expansion is to take place, and means being provided to increase the scanning frequency between the filter and the branching-off of the measurement path.

   9 5. Signal processing means as claimed in claim 4, comprising means to supply signals to the filter with a scanning frequency equal to or lower or higher than said normal scanning frequency.

   6. Signal processing means as claimed in claim 4, comprising means to cause the scanning frequency of the signals conducted to the regulating circuit to be increased in the measurement path by comparison with the scanning frequency used in the signal path, and to reduce said increased scanning frequency back to the scanning frequency in the signal path at said output of the measurement path.

   7. Signal processing means as claimed in any one of claims 3 to 6, comprising a decimation filter to effect reduction of the scanning frequency.

   8. Signal processing means as claimed in any one of claims 3 to 7, comprising an interpolation filter to effect increase of the scanning 15 frequency.

   9. Signal processing means substantially as hereinbefore described with reference to any one of Figs. 3 to 6 of the accompanying drawings.

   1 Published 1990 atThe Patent Mce, State House.66 71 High Holborn. London WC1R 4TP. Purther copies maybe obtamedfrom The Patent Otrice. EWes Branch, St Mary Cray. Orpington, Kent BR5 M. Printed by muitipiex techniques ltd, St Mary Cray. Kent, Con. 1187 1