The SPECCON : Spectral conversion of harmonics
The Speccon is a method and a device for generating higher harmonics.
It delivers a fully and bright sound according to an optimizing of psycho-acoustic data.
It works as well for speech as for music and has an automatic adjustable control.
The Speccon permits harmonics to be added f.i. for piano and trumpet.
For the piano it will be minimized and for trumpet enhanced, since the information signals
used for determining the extent to which the additionally created harmonics are added to
the output signal, can be chosen independently of the signals used for creating the harmonics itself.
For a good understanding of the Speccon, it is important to know what is the meaning of :
"generating higher harmonics"
An important step of the method according the invention consists in that higher harmonics
are generated starting from the isolated signals (5).
In general, in the field of processing audio signals, more particularly music,
for a man skilled in the art, the "generating of higher harmonics" means that the fundamental
tones are defined and that in function thereof a series of corresponding harmonics
(2nd, 3th, 4th, etc) are created (6).
In practice, the doubling or squaring of frequencies is not considered as a
"generating of higher harmonics" for the following reasons:
1. When squaring signals, these signals are in fact multiplied, which means
that starting from one signal another signal is produced having a higher amplitude and frequency.
Each input signal which is squared only results in one output signal whereby the input signal is lost.
This cannot be compared with the generating of higher harmonics
according to which the original harmonic is maintained and a complete
series of subsequent higher harmonics is generated.
2. When squaring the signals started from one input signal or a number of input signals
for example having the main frequency fH1.
When staring from one input signal it is not certain that this is a fundamental tone and it
may be that it is already a particular harmonic and which results that by squaring only a
further higher harmonic is generated, but no other harmonics are generated.
By processing the signals by means of squaring or multiplication, different results
are obtained because the tones are not enriched but only higher tones are created
resulting in special effects comparable with voices at high tones used in cartoons (Donald Duck-effect).
In a squaring circuit the input signal is multiplied by itself and the signal components are
generated having an amplitude which increases exponentially with the strength of the output
signal which results in a deformation of the sound.
A rectifier or a amplitude modulating circuit is often used to counteract this deformation
but in practice it does not function because the attack and delay time results
in changements in the signal which causes overshoot-effects.
It is also impossible to realize band pass filters providing a signal/noise-ratio
which is the square of the commonly used signal/noise-ratio in audio systems.
The separated and squared signals are additionally filtered in the band pass filters 29.
Due to the fact that these band pass filters 29 have to treat squared signals,
these band pass filters must have a signal/noise-ratio which is the square of the commonly
used signal/noise-ratio in audio systems.
If not, it is not possible to maintain the same quality of sound.
In reality it is impossible to realize such band pass filters providing a signal/noise-ratio of that high quality.
The present invention does not have these problems.
Most devices introduce serious deformations of the sound when polyphone signals are applied to the latter.
Therefore, in the case of music sound, which comprises a large plurality of polyphonic sounds,
those systems cannot be used in the reality.
In the Speccon, the present invention, the signals to be processed are not necessary divided
in two separated groups of frequencies, contrary to the invention described in US Patent 5.388.159.
According to US 5.388.159, there is a clear division in two separate groups, namely the first group of
low frequencies named fL and a second group of higher frequencies named fH1 to fH4. Due to this division
in separate groups, as well as for other reasons, the system disclosed in US 5.388.159 does not provide in an "enrichment"
of the signals to be treated, but, on the contrary, this prior art system provides a signal which is very poor with
regard to harmonics available in the signal.
The system disclosed in US Patent 5.388.159 does not provide for an enrichment with harmonics for the following reasons:
1. First Reason:
To obtain that the system disclosed in US Patent 5.388.159 can function, it is clear that the frequencies fH1
to fH4 have to be chosen such that they are sufficiently separated from each other.
This implicates that the total band width covered by the frequencies fH1 to fH4 is quite large and consequently
also implicates that the width of the remaining band fL is quite limited.
This means that the signal EL which is derived from the remaining ban fL and which is passed directly from the
filter device 22 to the adding circuit 32, can only comprise a very limited portion of the input signal 21.
As the signal EL only comprises a very limited portion of the input signal, it is also clear that only a very limited
number of original harmonics of the input signal, will be available in the signal EL.
Consequently, via the signal EL almost none of the original harmonics will arrive at the
adding circuit 32. Furthermore, via the line followed by signal EL there is no generation of additional harmonics.
Conclusion A: via the signal line of EL, practically no harmonics arrive at the input.
2. Second Reason:
The filter device 22 at the entrance provides for the division in frequency bands fH1 to fH4.
From the curves represented in block 22 of figure 10 or figure 14, it is clear that these frequency
bands are defined around main frequencies fH1, fH2, fH3 and fH4, whereby the main frequencies can optimally pass,
whereas the frequencies in between the main frequencies only pass to a limited extent.
This means that from the original harmonics available in the frequency bands fH1 to fH4,
only these harmonics from the frequencies higher than frequency fL are passed to the squaring circuits.
Conclusion B: only a limited number of original harmonics fro the frequencies higher than frequency fL are passed to the squaring circuits.
3. Third Reason
The signals EH1 to EH4 are passed to the squaring circuits 28-1 to 28-4. As explained,
in these squaring circuits the input signals are squared, which means that the main frequencies are doubled.
The original frequencies however do not pass and only the doubled frequencies do pass.
This means that original harmonics which pass through the filter device and arrive at the squaring circuits,
are removed and that only the square signal thereof is further passed to the adding circuit.
This also means that the original harmonics having a frequency which is higher than fL (and most of them will be higher than fL), are lost.
Conclusion C: all original harmonics reaching the squaring circuits are lost.
4. Fourth Reason
The simple "squaring of signals" cannot be compared with "generating higher harmonics" (see explanation above).
Conclusion D: even in the case that an original harmonic arrives at the entrance of a squaring circuit,
it is clear that only one result is created, which is one particular higher harmonic of the original harmonic.
This cannot be compared with the classical series of higher harmonics which is created when "generating higher harmonics" in the common manner.
From the conclusions A, B, C, and D; it is clear that no or almost no higher harmonics will arrive at the adding circuit.
Consequently, there is certainly no "enrichment" of harmonics as thought by the invention.
On the contrary, the output signal of the US Patent 5.388.159 is very poor with regard to harmonics available in the obtained signal.
General Conclusion
That the system of US Patent 5.388.159 cannot be used to ameliorate the sound of music can be illustrated by the following example.
When an input signal is presented to the system having a frequency higher than fL, due to the squaring circuits,
it is clear that the fundamental tone gets lost. After the squaring circuits, in which both amplitude and frequency are squared,
it is clear that only the higher upper tones remain. Moreover, due to the fact that there is a doubling of the frequency, it is clear
that the normal ration between the upper tones also gets lost. As a result, a normal tone of high frequency is transformed
in a robot-like "Donald Duck" voice or tone.
Further more, very high tones of the last frequency band fH4, are removed completely.
In fact, as explained before, the basic tones themselves do not pass the corresponding square filter.
The squared value of this signal does pass, however so high that it is not perceptible.
Our present invention does provide for a solution for this.
The band filters 13 and 20 can be chosen independently fro each other, which is not possible
in the system disclosed by the cited patent. According to this prior art, there is a clear division
in frequency bands and a particular choice of bands fH1 to fH4, automatically results in a particular band fL.
Consequently, according to this prior art, the frequency band comprising frequencies fH1 to fH4 can never overlap
with the frequency band fL. According to the present invention the band widths of the filters 13 and 20 can be chosen
in such a manner that the ranges of frequencies overlap each other. In this way an optimal choice can be made in
function of the application of the system.
