COMMISSIONER'S DECISION
Computer Related Subject Matter: Determining Pitch in Human Speech
The system provides components for deriving pitch values of signals,
for preparing masks for passage of derived values, and for determining
from the values passed through the masks a match to enable voice
signals to be transmitted. Rejection of application withdrawn.
This decision deals with Applicant's request for review by the Commissioner of
Patents of the Final Action on application 341,411 (Class 354-53) filed
December 6, 1979. It is assigned to N.V. Philips Gloeilampenfabrieken and is
entitled METHOD OF AND SYSTEM FOR DETERMINING THE PITCH IN HUMAN SPEECH. The
inventors are H. Duifhuis, L.F. Willems, R.J. Sluyter. The Examiner in charge
issued a Final Action on June 15, 1982 refusing to allow the application.
This application relates to a system for identifying whether incoming speech
signals, for example originating from telephone lines, match stored
predetermined speech signals. Predetermined pitch values are derived for peak
positions of the amplitude spectrum of a stored speech signal, as well as for
the intervals around the values. Two separate processes are disclosed for
comparing the pitch of a transmitted signal with the predetermined values by
means of preparing a reference mask. One process forms a mask with apertures
in predetermined places, indicative of the value for the pitch and multiples
thereof. The other makes a mask which has apertures representing predetermined
significant peak portions. Each process acts on incoming signals by computing
their particular values, subjecting the values to passage through its mask,
ascertaining the values of the signals which pass, and determining whether the
values of the signals so passed match the values of the stored signals.
Figures 4A and 4B reproduced below, illustrate apparatus used in the system:
<IMG>
Elements 101 through 107 process an input signal 100 into certain computed
values and 108 stores them. A pulse generator 109 via line 127 provides
signals to micro-computer 114 while line 125 leads the computed values to
central processing unit (CPU) 112. Thereafter, interaction of the elements in
block 111 perform the steps of the processes to produce the matched values of
the speech signals as output signals at output register 116. Each of the
processes is said to be the dual of the other.
In refusing the application for being directed to non-statutory subject matter,
the Examiner reasoned in part, as follows:
. . .
... It is not evident that the claimed process uses
apparatus other than general purpose computing apparatus.
The apparatus listed on page 2 of the letter of March 8,
1982 appears to be conventional apparatus for preparing
speech signals for processing by a data processor (i.e. it
is conventional to convert the analog speech signal to
digital form to enable processing by a digital computer).
. . .
... Thus the applicant must show that the claimed process is
carried out on novel apparatus wherein a computer may be a
component but wherein the novelty lies in a combination of
which the computer is only one component. The novelty may
not lie is the software in the computer as appears to be the
case herein.
. . .
... Applicant further states on page 2 of the letter that a
system could be hardwired to carry out the method in
accordance with the invention. It would be necessary to
disclose such hardware in accordance with Section 36(1) of
the Patent Act if this argument were to carry any weight.
It is held that the process of Fig. 1 has only been dis-
closed as flow charts of a program in Figs. 2, 3, 5 and that
the apparatus disclosed to carry out the process shown in
Fig. 4 fails to disclose novel apparatus. The apparatus in
Fig. 4 is conventional general purpose computing apparatus
in combination with expected auxillary apparatus such as a
code converter, buffer, etc.
The Applicant presented his arguments for allowance of the application, in
part, as follows:
... The manner in which the known components are used to
process the speech signal to a state suitable for input to a
general-purpose computer, programmed in a special manner, is
a part of the complete system of speech analysis as claimed
in the present application. ... line 36 of page 6 clearly
indicates that the functions of blocks 25 onward is imple-
mented by the software of a general-purpose computer, the
software mentioned herein directing operations additional to
those performed by blocks 22-25.
... the first paragraph of the disclosure on page 7 ...
states "By way of input data the computer receives the com-
ponents AF(r), r = 1, ...., 128 of the amplitude spectrum as
represented by block 26". This paragraph ... clearly indi-
cates that the complete system of speech analysis requires
more than a specifically programmed general-purpose computer
in order to produce the data input for such computer.
. . .
The processing of the speech to data, suitable for input to
the computer, is necessarily carried out by apparatus or
hardware which may be that according to Figure 4 as des-
cribed beginning at line 18 of page 13.
The hardware of Figure 4 comprises a low-pass filter 101, a
sampling switch 102, an analog-to-digital converter 103, a
buffer store 104, a multiplier 105 for discrete Fourier
transformation based on coefficients supplied by ROM 106
producing "frequency points" and a buffer store 108.
It should be readily apparent that a low-pass filter, such
as 101, is not a component part of a general-purpose com-
puter, nor is it possible to feed other than so-called data
signal (pulse signals) to a general-purpose computer. In
the present system, the input to filter 101 is an analog
speech signal.
Furthermore, for the Fourier transform operation of block
105, a ROM 106 is specially provided with the coefficients
necessary for the particular operation desired.
The output of accumulator 108 is clocked as input to the
general-purpose computer, block 111 of Fig. 4B again giving
clear indication that the general-purpose computer is con-
cerned only with the final mathematical manipulations of the
data resulting from the processing of the analog speech
signal.
. . .
It is only possible to select the time segments and deter-
mine Fourier transforms when the analog speech signal is
processed to a form of selected time segments exhibiting a
Fourier transform characteristic. This part of the system
or method of speech analysis is performed by apparatus other
than a general-purpose computer and hence the complete sys-
tem, as claimed, including the selecting, Fourier trans-
formation and final mathematical analysis, the last part of
which may be performed by a general-purpose computer.
The issue before the Board is whether or not the application is directed to
statutory subject matter in view of Section 2 of the Patent Act. Claim 1
reads:
In a system of speech analysis wherein the amplitude
spectrum of a speech signal is analyzed by regularly select-
ing time segments of the speech signal, by determining from
each time segment a sequence of spectrum components which
constitute the discrete Fourier transform of samples of the
speech signal and by deriving in each time segment the posi-
tions of the significant peaks in the spectrum from the
sequence of spectrum components, the method comprising the
steps:
the selection of a value for the pitch and the determination
of s sequence of consecutive integral multiples of this
value and the determination of intervals around this value
and the multiples thereof, these intervals defining a mask
having apertures in situ of an interval, harmonic number
corresponding to the multiplication factors in the said
multiples being associated with the apertures;
the determination of the significant peak positions coincid-
ing with a mask aperture;
the computation of a quality figure in accordance with a
criterion indicating the degree to which the significant
peak positions and the mask apertures match;
the repetition of the preceding steps for consecutive higher
values of the pitch until a predetermined highest value,
resulting in a sequence of quality figures associated with
these pitch values;
the selection of the value of the pitch having the highest
quality figure, of which the associated mask constitutes a
reference mask;
the association of the harmonic numbers of the apertures of
the reference mask with the significant peak positions
coinciding with the apertures, these harmonic numbers char-
acterizing the locations of these peak positions in a
sequence of harmonics of a same fundamental tone; and
the determination of a probable value for the pitch, thus
that the deviations between the last-mentioned significant
peak positions and the corresponding multiples of the prob-
able value having the same harmonic numbers are as small as
possible.
We look to the decision in Schlumberger Canada Ltd. v. The Commissioner of
Patents (1981) 56 CPR (2d) at 204 in determining whether the application is
directed to statutory subject matter, and in paticular to the following
passages of Pratte, J.:
In order to determine whether the application discloses a
patentable invention, it is first necessary to determine
what, according to the application, has been discovered.
and
I am of opinion that the fact a computer is or should be
used to implement discovery does not change the nature of
that discovery. What the appellant claims as an invention
here is merely the discovery that by making certain calcula-
tions according to certain formulae, useful information
could be extracted from certain measurements. This is not,
in my view, an invention within the meaning of Section 2.
Applicant has described an arrangement which identifies transmitted speech
signals by determining pitch values of those signals and comparing them with
stored predetermined pitch values of speech signals. The system provides
various means for deriving pitch values of received signals, preparing masks
for passage of the derived values, and determining from the values passed
whether or not a match of pitch values is present. In our view, the system
provides s useful end result in matching acceptable voice signals that are
transmitted from one place to another, for example, in telephone communica-
tion. We consider the matching of the signals is directed to more than merely
making calculations. We are satisfied therefore that the application is
directed to patentable subject matter.
We note the claims define a system to analyse speech signals by regularly
selected time segments, and by determining from each segment a sequence of
spectrum components. We see that they set forth the various steps which
determine the pitch values of speech signals, provide appropriate reference
mask means, and determine the correspondence of incoming signals to stored
values. In our view, they are directed to the invention disclosed.
In summary, we are satisfied the application discloses and claims features that
present more than calculations to convert a set of values into another set of
values. We are persuaded that the application and claims are directed to
allowable subject matter, and in the absence of any cited art, may be allow-
able.
We recommend that the rejection of the application for being directed to
non-statutory subject matter, be withdrawn.
M.G. Brown S.D. Kot
Acting Chairman Member
Patent Appeal Board
I concur with the findings and recommendations of the Patent Appeal Board.
Accordingly, I withdraw the Final Action, and I am remanding the application to
the Examiner for prosecution consistent with the recommendation.
J.H.A. Gari‚py
Commissioner of Patents
Dated at Hull, Quebec
this 14th day of August 1986.
C.E. Van Steinburg
Philips Electronics Ltd.
601 Milner Ave.
Scarborough, Ontario
M1B 1M8