Signal Identification System

Abstract

A system is disclosed for reducing analog signal patterns or sequences to concise digital representations for identification, as to recognize an audio signal, for example, manifesting a musical recording. In the system, audio (analog) signals are sampled for dissection into a plurality of values which are accumulated over a sampling interval to provide an aggregate value. Preliminary to such a sampling operation, the signal of interest may be frequency dissected into a plurality of individual signals which are sampled then summarized or totalled. The aggregate total numerical values are reduced to a representative form by encoding them in accordance with their relative significance. Operation over repeated sampling intervals affords an expanded basis for indentification and recognition.

Parent Case Text

CROSS REFERENCE TO RELATED APPLICATIONS

This application is a continuation of application Ser. No. 46,352, filed June 15, 1970 and now abandoned.

Description

BACKGROUND AND SUMMARY OF THE INVENTION

The desire sometimes occurs to recognize, or identify analog signal patterns, as for example, an interval of an audio signal. As a specific example, it may be desirable to monitor a radio broadcasting station with the objective of providing information on the individual musical selections that are broadcast by the station during a particular logging interval. That is, rather than to manually record a broacast log as a basis for royalty payments, and so on, a signal-recognition system could develop a compilation of the broadcast material with improved economy and accuracy.

It has been previously proposed to monitor a broadcast station by providing a code on each and every recording that is to be identified if and when it is broadcasted, then detecting the occurrence of the code in the broadcast radio signal, as a basis for compiling a broadcast log. Although such an arrangement affords certain advantages, its commercial implementation has appeared to be exceedingly difficult. Specifically, a great problem is presented in obtaining and maintaining recorded codes on all the recordings of interest which may be contained in the broadcast material from a station. Additionally, broadcasting codes and so on may present difficulties with regulatory authorities and furthermore, substantial difficulty arises with regard to the problem of selectively sensing the coded information, i.e., discriminating in favor of code signals and against the other content of a broadcast signal. Consequently, systems of this type have not generally been considered for serious commercial use.

In any signal-recognition system, for use to identify a particular sequence or pattern that is contained in a substantially-continuous signal, one of the problems is related to timing. That is, if a recognition system were provided with synchronizing signals as well as the signal of interest, the problem would be substantially simplified. However, in the exemplary application considered above, i.e., monitoring the signal from a commercial radio broadcasting station, synchronizing information is not present to indicate intervals of concern. Accordingly, a system for such an operation must have the capability to monitor a substantially-continuous signal, e.g., the audio content of a standard radio broadcast signal, and recognize the occurrence of specific sequences or patterns, e.g., recordings, contained within the broadcast signal, without the benefit of any timing or synchronizing signals from the radio station.

In general, the system hereof functions to accomplish the objectives considered above by developing digital numerical representations that are indicative of sequences of the signal under consideration. Such numerical representations are developed by dissecting the analog signal of interest (sampling, filtering, and encoding). The system may be implemented to search for a coincidence with any of many similarly-developed numerical representations which are established in staggered relationship to each other.

BRIEF DESCRIPTION OF THE DRAWINGS

In the drawings, which constitute a part of this specification, exemplary embodiments demonstrating various objectives and features hereof are set forth as follows:

FIG. 1 is a block diagram of an exemplary system incorporating the structure of the present invention.

FIGS. 2 and 3 are sets of waveforms illustrative of the operation of the present invention;

FIG. 4 is a block diagram of one embodiment constructed in accordance with the present invention;

FIG. 5 is a block diagram of a portion of the system of FIG. 4;

FIG. 6 is a diagram illustrating the operation of the system of FIG. 4; and

FIG. 7 is a block diagram of another embodiment incorporating the principles of the present invention.

DESCRIPTION OF THE ILLUSTRATIVE EMBODIMENTS

Referring initially to FIG. 1, a system is generally depicted for monitoring the broadcast content of a plurality of radio stations by using multiplexing techniques. Specifically, a plurality of radio receivers R1 through RN are represented, each of which is tuned to receive a selected-station radio broadcast signal. Of course, the number of individual radio receivers provided depends upon the number of stations desired to be monitored and the extent of multiplexing.

Each of the radio receivers R1 through RN provides a detected audio signal to a signal pattern recognition system 12, which reduces each of the received audio signals (as described below) to a series of digital representations. The operation is similar with regard to each of the received audio signals, the analysis of each resulting in separate and distinct data. Accordingly, one signal may be considered as representative of all for purposes of explanation. Therefore, assume, for example, that a particular recording (musical selection) is transmitted by the station monitored by the receiver R1. That audio signal (representing a musical selection) is reduced to a sequence of digital signal representations that are passed through a cable 14 to a storage unit 16. Specifically, for example, (each two seconds) of the audio signal under observation might be analyzed or dissected to provide one set of digital representations, e.g., a seven-digit decimal numeral. Accordingly, a particular recording can be productive of a series of numerical values, each being developed from a brief interval of the recording.

A pattern of such numerals may thus identify a particular sequence of an audio signal. Consequently, as disclosed in detail below, such data may be cataloged in advance, for comparative analysis so as to recognize that a specific recording has been broadcast. Thus, the significant consideration is that, a sequence of an analog signal extending over a time interval and with almost infinite possibilities, is reduced to a concentrated set of digital values. Furthermore, the digital values may be repeatedly duplicated and consequently truly identify the analog sequence.

It is to be appreciated that the digital numerals as set forth above (developed from a specific recording) may vary depending upon the precise instant of the starting interval during which the audio signal is analyzed or dissected. That is, if the first two seconds were at the very beginning of a recording, the numerals may differ substantially from numerals that would be developed if the interval was started one half second into the recording. As a consequence, in applications of the system hereof, several thousand time-displaced, or staggered, sets of identifier numerals may be developed and used for each recording which is to be identified by the system. That is, it is to be appreciated that a great many sets of numerals may be cataloged (as on a template tape) to identify one recording. Comparisons may then be performed, in accordance with somewhat conventional techniques to indicate identification.

In an exemplary system hereof, a catalog of digital numerals for records of concern is registered by a template tape unit 18 and signals representative of the digital numerals are supplied to a comparison unit 20 as indicated. The comparison unit 20 also receives the signal-represented numerals from the storage unit 16, either directly or indirectly as indicated by a dashed line 22. The comparison unit manifests the coincidence of similar sets of numerals contained in the data from the template tape unit 18 and that from the storage unit 16. Upon the occurrence of a coincidence, a particular recording is designated whereupon a signal identifying that recording is passed through a cable 24 to a recording unit 26 as the basis for a log of the broadcast monitored by the radio receiver R1.

The system as disclosed in FIG. 1 is indicated to operate in cooperation with a plurality of radio receivers R1 through RN. As suggested above, although description has been directed to a single audio signal, it is to be appreciated that by the use of well-known time-sharing and multiplex techniques, a plurality of broadcast stations can be concurrently monitored by means well knwon in the art. In that regard, the signal pattern system 12 comprises the key element of the combination.

In view of the above preliminary description of the present system, reference will now be made to FIG. 2 (and subsequently to FIG. 3) for an analytical consideration of the techniques for developing sets of numerals from sequences of an analog or audio signal. Specifically, the curve of FIG. 2(a) represents an audio signal that is to be defined over a relatively-short interval of time, e.g., two seconds. The signal represented in the curve of FIG. 2(a) varies in amplitude about a reference level indicated by a line 30. As an initial step, the signal as represented may be rectified with reference to the reference level to provide a single polarity signal as represented by the curve 32 of FIG. 2(b).

One of the definitive characteristics of the curve 32 is the total area thereunder. That is, an indication of the total area under the curve 32 during the interval T is an identifying characteristic of the curve. One technique for developing a representative value for the area under the curve is to sample the signal represented by the curve 32 at regular intervals and accumulate the aggregate value of the individual samples. Such an operation may be related to the mathematical process of integration. That is, the curve 32 may be dissected into a plurality of sample amplitude values as indicated by the lines 34. By accumulating such values represented by the lines 34 over the interval T, an aggregate value is developed which is related to the area defined under the curve 32.

Although the area under a curve is an identifying characteristic of a curve, it is rather ambiguous in that many totally-different curves will define similar areas. Consequently, further identification of a signal, as represented by a curve, is generally necessary in a practical signal-identification system.

Prior to dissecting a signal by sampling, (as considered above) the curve may be dissected on a frequency basis. For example, an audio signal occurring during a time interval T might be filtered to provide three component frequency signals which may then be rectified to forms as represented by the curves shown in FIGS. 3(a), 3(b) and 3(c). By frequency dissecting an analog signal, its unique characteristics are more effectively defined by three separate signals (FIG. 3) which then may be individually sampled to develop three aggregate numerical values. Such values are considerably more definitive of the signal than the prior single value as considered above with reference to FIG. 2. Thus, if each of the signals represented by the curves of FIGS. 3(a), 3(b) and 3(c) are individually sampled over an interval T and the samples for each signal are accumulated, three values of considerable significance and definition will result.

In accordance herewith, the aggregate numerical values developed by the method considered above are effectively reduced by encoding them on the basis of their relative significance. That is, an identifying code value is assigned to the analog signal, dependent upon the relative magnitudes of a plurality of numerical values developed, which code affords a rather concise identification for the signal.

Recapitulating, it may be seen that the system hereof is disclosed to utilize both frequency dissection and sampling dissection to develop numerical values which are then encoded to more concise digital forms based upon relative significance. The resulting digital codes then afford concise digital identification of a sequence of an analog signal.

Although the above analytical explanation was based on a frequency dissection of the analog signal into three separate signals, it is readily apparent that various numbers of frequency bands may be employed, depending upon the degree to which a signal is to be defined. Somewhat similarly, sampling rates and coding formats may vary within wide limits depending upon the peculiar demands imposed upon a particular system.

A system for dissecting an analog signal somewhat as considered above with reference to FIG. 3 is shown in FIG. 4 and will now be considered in detail. The analog signal is applied to a terminal 36 which is connected to the input of a multiple band-pass frequency filter 38. Various forms of filter devices may be employed as the filter 38 to provide three distinct signals in conductors 40, 42 and 44 respectively, which manifest three selected-frequency components of the received analog signal. In one form, the filter 38 may comprise simply an amplifier with an output connected in series to three tuned circuits which have transformer-coupled outputs, as well known in the prior art of frequency division.

The conductors 40, 42 and 44 receiving the three selected frequencies are connected respectively to sampling circuits 46, 48 and 50 which are also connected to receive pulses from a clock generator 52 which may take various forms of that structure as well known in the prior art. The sampling circuits 46, 48 and 50 may comprise forms of "and" logic gates which are enabled upon receipt of a clock pulse to briefly pass the instant analog signal amplitude received from the filter 38.

The analog values provided from the sampling circuits 46, 48 and 50 are applied to accumulators 56, 58 and 60 respectively. These structures may simply comprise analog registers, as well known in the prior art, some of which accumulate a potential as a capacitive charge to manifest an accumulated analog value. Of course, various forms of analog accumulators are well known in the art.

The accumulators 56, 58 and 60 are connected respectively to gate circuits 66, 68 and 70 which also each receive a qualifying signal from a counter 72 that is driven by the output from the clock circuit 52. When the counter 72 reaches capacity, the next pulse received thereby causes an overflow from the counter, which qualifies the gates 66, 68 and 70 resulting in the application of the contents of the accumulators 56, 58 and 60 to a digital coding unit 74.

As will be described in detail below, the digital coding unit 74 (here exemplified in a single form) provides binary-value signals through a gate 76 to a register 78 which are developed to represent a value in accordance with the relative magnitudes of the signals received from the gates 66, 68 and 70. The binary code value thus identifies an interval of the analog signal which is received at the terminal 36 from which such value was developed.

Assume now, for example, that an operating interval of the system is defined to be two seconds, and that at the conclusion of each such two-second interval, the system provides a representative binary number to the register 78. Assume further, as indicated above, that the analog signal applied at the terminal 36 is dissected to provide three component-frequency signals which are sampled at the rate of 100 samples per second. Accordingly, the clock 52 provides pulses at 1/100th of a second and the counter 72 has a capacity to tally the clock pulses to a count of 200, so that the counter overlfows every two seconds.

Assume still further as an operating aspect, that the filter 38 provides fequency component signals as follows: (1) output to conductor 40: 500 cycles; (2) output to conductor 42: 1,000 cycles; and (3) output to conductor 44: 2,000 cycles. Accordingly, every 100th of a second, an analog value is supplied from each of the sampling circuits 46, 48 and 50 to be added to the contents of an associated one of the accumulators 56, 58 and 60 respectively. Additionally, after each two second interval of accumulation, the contents of the accumulators 56, 58 and 60 are gated in the form of signals .SIGMA.1, .SIGMA.2, and .SIGMA.3, through gates 66, 68 and 70 respectively to the digital coding unit 74. At the instant concluding a two-second interval, the digital coding unit 74 receives the overflow signal from the counter 72 commanding a "test" operation whereby the order of significance of the signals .SIGMA.1, .SIGMA.2, and .SIGMA.3 is determined as a unique situation to provide code values that are supplied from the digital coding unit 74 through the gate 76 to the register 78. A very brief interval after the digital coding unit 74 is actuated by the overflow pulse (applied at a "test" input) the unit is reset or cleared by receipt of a pulse at a "clear" input which pulse is received from the counter 72 through a delay circuit 77.

The digital code developed by the coding unit 74 may follow any of a great many different formats depending upon specific application requirements and design. For example, the digital-value signals may be derived on the basis of the relative significance of the values accumulated in the accumulators 56, 58 and 60, somewhat as described hereinafter. Alternatively, various other coding formats can be employed as will be apparent to those skilled in the art.

A fragment of one form of digital coding unit 74 is disclosed in detail in FIG. 5. As shown, the signals .SIGMA. 1 and .SIGMA. 2 are applied to a comparator 80 (in a " .SIGMA. 1 >.SIGMA. 2" signal source 81) which comparator may take the form of an analog subtractor. In the event that the signal .SIGMA.1 exccces the value of the signal .SIGMA.2, an output is provided from the comparator to an "and" gate 82 which is qualified by the "test" signal and which sets a flip flop 84 indicating relative magnitudes of the two signal-represented values. As a consequence, a binary signal ".SIGMA.1>.SIGMA. 2" (descriptively designated) from the flip flop 84 becomes high and the negation thereof becomes low. The binary signal ".SIGMA.2>.SIGMA.3" (descriptively designated) may be developed by a source 87 similar to the source 81.

The signals ".SIGMA.1>.SIGMA.2" and ".SIGMA.2>.SIGMA.3" along with similar signals may then be applied to logic "and" gates as well known in the art to develop digital signal codes according to a selected format as described in greater detail below.

Recapitulating, the system of FIG. 4 receives an analog signal and after a predetermined test interval (two seconds) provides a binary value representative of the signal during the two-second interval. Consequently, a series or sequential set of such binary numbers may be effectively employed to recognize a specific analog signal. That is, having once analyzed a particular pattern of an analog signal as from an audio recording, representative digital signals may be developed for comparison with a continuous analog signal (which may embody the recording) in search of a coincidence to indicate similarity.

In a specific exemplary application hereof, as suggested above, a broadcast radio station might be monitored to obtain a continual compilation of digital representations including those for musical selections, i.e., recordings broadcast from the station. Of course, to selectively determine the presence of digital representations indicative of a recording and to identify each specific recording it is necessary to construct a catalog or library of digital representations that identify the recordings of concern. Accordingly, such recordings are analyzed in advance and representative sets of digital values are developed which identify each recording. In view of the fact that the signal under observation, as from a radio broadcasting station, is not time synchronized, it is necessary to develop several thousand sets of numerical sequences (indicative of a recording) which are time-displaced in overlapped relationship. For example, the observed set of digital values for a specific recording (musical selection) may vary with the time relationship of the encoding system to the record signal. Accordingly, a plurality of time displaced sets of digital values are developed so that coincidence with any one of the plurality during a monitoring operation indicates that the recording in question is contained in the observed signal. Of course, varying standards of identify and deviation may be employed depending upon the nature of the system.

In an exemplary form hereof, an identification for a recording may consist of ten digital values (derived from ten two-second intervals). In such a format, the catalog of template identifications for each of the recordings may be forty sets of digital values. For example, assume that analysis and encoding of an analog signal is initiated at a time t1 and that analysis and encoding continues throughout a twenty second interval, to provide one set of digital values for the template catalog. A short interval after the time t1, e.g., 1/25th of a second later, at a time instant t2, another analysis is initiated to formulate still another set of digital values for a second identification. Thus, in an exemplary form hereof, some 40 seconds of analog signal is employed to produce 20 sets of digital values which are offset by 1/25th of a second and each of which includes 20 seconds of the analog signal. The format of this operation is illustrated diagrammatically in FIG. 6. At the time instant t1, a 20 second interval of play is initiated from which one set of digital values are developed. At time t2 (one 25th of a second after t1) another twenty second time interval is initiated during which another set of digital values is developed. Accordingly, twenty sets (No. 1 through No. 20) of digital values are developed for comparison with the continuous flow of digital identifications that are developed from the monitored signal during its observation. Upon a coincidence, or near coincidence, of any set with a sequence from the monitored signal (as provided by the structure as depicted in FIG. 4) a recording is identified.

Recapitulating, a system as disclosed in FIG. 4 is first used to obtain several thousand sets of catalog digital values which are developed from time-displaced portions of an analog signal as may be provided from a musical recording. Thereafter, a continuous analog signal (including the signal forms of the musical recordings) is monitored to provide a sequence of digital values, using a system as shown in FIG. 4. The catalog values and the analog signal values are then compared in a data processing operation to detect the appearance of specific recordings in the analog signal. Thus, a radio broadcasting station can be monitored (internally or externally) to provide a tabulation of the specific recordings which are broadcast, without the necessity of laborious and often inaccurate manual monitoring procedures.

As described above, the system is substantially analog in form. However, it may be advantageous to provide the system in a form which is more extensively digital in nature. Such a system is shown in FIG. 7 and will now be considered in detail to set forth another embodiment hereof as a basis for the claims.

Again, the system is disclosed to include a sampling unit for dissecting an analog input. Specifically, the received audio or other analog signal is supplied through a conductor 102 to a signal sampling unit 104 which is synchronized by pulses that are received from a clock generator 106. Of course, the sampling rate may vary widely and in accordance with current technical capabilities, however, sampling rates of 200 per second are reasonable in a present embodiment of the system.

The sampled signals (defining the input analog signal) are supplied from the sampling unit 104 to an analog-to-digital converter 108 that is also synchronized by the signal from the clock generator 106. Various forms of such converters are well known in the prior art which are capable of supplying digitized representations of analog samples at a predetermined sampling rate.

The digital representations (series or parallel) from the converter 108 are supplied through a cable 110 to a multiple band pass digital filter 112. The digital filter 112 provides four outputs, (1), (2), (3) and (4), each of which is a set of signal represented digital values for four specific frequency bands of the analog signal received through the cable 110. That is, the digital filter 112 is a digital equivalent of the well-known analog, multiple-band-pass filter apparatus. A form of digital filter suitable for use herein is disclosed in a publication entitled "An Approach to the Implementation of Digital Filters" , by Jackson, Kaiser and McDonald; published in IEEE Transactions on Audio and Electroacoustics; Volume AU-16, Number 3, September 1968.

The four sets of digital signals (1), (2), (3) and (4) from the digital filter 112 each consist of a series of digital values definitive of the amplitude of one frequency component of the original analog signal under investigation. The signals (1), (2), (3) and (4) are applied respectively to digital accumulators 114, 115, 116 and 117. The signals, as accepted by the accumulators, are representative of absolute values. That is, the sign digit of the signal or value represented is dropped. This consideration is related to the rectification of component signals in the analog form of the present invention.

The accumulators 114, 115, 116 and 117 may take a variety of different forms, normally including a digital adder and a register interconnected as well known in the prior art. In addition to the digital signals, the accumulators each receive a binary "clear" signal from a counter 120 which is connected to be driven by pulses from the clock generator 106. If, for example, the clock generator operates at a rate of 200 pulses per second, the counter 120 may have a capability to count to 400 so as to provide the accumulations of a two second accumulation interval.

When the capacity of the counter 120 is exceeded, an overflow pulse is provided to clear the accumulators 114, 115 116 and 117, and simultaneously transfer their contents through a set of gang "and" gates 124, 125, 126 and 127 respectively to a group of comparator circuits 130, 141, 142, 144, 146 and 148. The gang "and" gates may each consist of several individual "and" gates as well known in the prior art, associated with the individual conductors in the cables carrying the output signals from the accumulators 114, 115, 116 and 117.

The signals that represent the digital values which are passed through the gang gates 124 and 125 are applied to a comparator 130 which (as the other comparators) has two binary outputs. If the summation of the component value represented by the signal (1), i.e., .SIGMA.(1), exceeds the value indicated by the summation of the signals (2), i.e., .SIGMA. (2), a high signal will appear in the conductor 132. In the event that the inverse is true, a high signal appears in the conductor 134. This format is provided by each of the comparator circuits of FIG. 7, including comparator circuits 141, 142, 144 and 146 and 148.

The comparator circuits may comprise a form of binary subtractors as well known in the prior art, coupled to a flip flop circuit substantially as disclosed in FIG. 5. In the event for example, that the signal-represented subtraction of the value .SIGMA.(2) from the value .SIGMA. (1) results in a positive difference, a high level binary signal is received by the conductor 132. Conversely if the subtractive operation results in a negative value, the conductor 134 receives a high binary signal. The operation of the comparators 141, 142, 144, 146 and 148 is similar with the result that twelve binary inputs C1 through C12 are developed which are indicative of the relative magnitudes of the summation quantities provided from the accumulators 114, 115, 116 and 117, to a decoding matrix 150. The function of the decoding matrix is to perform logical operations to develop five-bit binary code signals controlled by the relative magnitudes of each of the summed quantities. Specifically, the relationship of the summed quantities to the logic combination of signals C1 through C12 and the resulting output codes are set forth in the following chart:

Relationship Logic Signals Code .SIGMA.(4) > .SIGMA.(3) > .SIGMA.(2) > .SIGMA.(1) C11.sup.. C3.sup.. C1 11111 .SIGMA.(4) > .SIGMA.(3) > .SIGMA.(1) > .SIGMA.(2) C11.sup.. C5.sup.. C2 11110 .SIGMA.(4) > .SIGMA.(2) > .SIGMA.(3) > .SIGMA.(1) C7.sup.. C4.sup.. C6 11101 .SIGMA.(4) > .SIGMA.(2) > .SIGMA.(1) > .SIGMA.(3) C7.sup.. C2.sup.. C6 11100 .SIGMA.(4) > .SIGMA.(1) > .SIGMA.(3) > .SIGMA.(2) C9.sup.. C6.sup.. C3 11011 .SIGMA.(4) > .SIGMA.(1) > .SIGMA.(2) > .SIGMA.(3) C9.sup.. C2.sup.. C4 11010 .SIGMA.(3) > .SIGMA.(4) > .SIGMA.(2) > .SIGMA.(1) C12.sup.. C7.sup.. C1 11001 .SIGMA.(3) > .SIGMA.(4) > .SIGMA.(1) > .SIGMA.(2) C12.sup.. C9.sup.. C2 11000 .SIGMA.(3) > .SIGMA.(2) > .SIGMA.(4) > .SIGMA.(1) C3.sup.. C8.sup.. C9 10111 .SIGMA.(3) > .SIGMA.(2) > .SIGMA.(1) > .SIGMA.(4) C3.sup.. C1.sup.. C10 10110 .SIGMA.(3) > .SIGMA.(1) > .SIGMA.(4) > .SIGMA.(2) C5.sup.. C10.sup.. C7 10101 .SIGMA.(3) > .SIGMA.(1) > .SIGMA.(2) > .SIGMA.(4) C5.sup.. C2.sup.. C8 10100 .SIGMA.(2) > .SIGMA.(4) > .SIGMA.(3) > .SIGMA.(1) C8.sup.. C11.sup.. C5 10011 .SIGMA.(2) > .SIGMA.(4) > .SIGMA.(1) > .SIGMA.(3) C8.sup.. C9.sup.. C6 10010 .SIGMA.(2) > .SIGMA.(3) > .SIGMA.(4) > .SIGMA.(1) C4.sup.. C12.sup.. C9 10001 .SIGMA.(2) > .SIGMA.(3) > .SIGMA.(1) > .SIGMA.(4) C4.sup.. C5.sup.. C10 10000 .SIGMA.(2) > .SIGMA.(1) > .SIGMA.(4) > .SIGMA.(3) C1.sup.. C10.sup.. C11 01111 .SIGMA.(2) > .SIGMA.(1) > .SIGMA.(3) > .SIGMA.(4) C1.sup.. C6.sup.. C12 01110 .SIGMA.(1) > .SIGMA.(4) > .SIGMA.(3) > .SIGMA. (2) C10.sup.. C11.sup.. C3 01101 .SIGMA.(1) > .SIGMA.(4) > .SIGMA.(2) >.SIGMA.(3) C10.sup.. C7.sup.. C4 01100 .SIGMA.(1) > .SIGMA.(3) > .SIGMA.(4) > .SIGMA.(2) C6.sup.. C12.sup.. C7 01011 .SIGMA.(1) > .SIGMA.(3) > .SIGMA.(2) > .SIGMA.(4) C6.sup.. C3.sup.. C8 01010 .SIGMA.(1) > .SIGMA.(2) > .SIGMA.(4) > .SIGMA.(3) C2.sup.. C8.sup.. C11 01001 .SIGMA.(1) > .SIGMA.(2) > .SIGMA.(3) > .SIGMA.(4 )C2.sup.. C4.sup.. C12 01000

from the above chart, it is apparent that a series of binary signals representative of five-bit binary numbers are developed at the output of the decoding matrix 150 which values are indicative of an interval of observation of the analog signal. As in the embodiment considered above, the interval may consist of two seconds or may be varied to accommodate other objectives. The structural details of the decoding matrix 150 are explicit from the above chart in that the matrix simply incorporates elements to accomplish the specific logic comparison, i.e., "and" combinations.

The sets of binary signals developed from an interval of the analog signal may be recorded as described above with reference to FIG. 4 and employed as a part of a template catalog for comparison with a minotored signal as described with reference to FIG. 1. That is, having obtained a digital representation resulting from the dissection of an analog signal for an interval, such digital representations may be variously employed as to identify portions of an analog signal under observation.

In some instances of operating a system in accordance herewith, the summation values may be too close to each other to draw a significant distinction. In such a situation, one of the comparators 130, 141, 142, 144, 146 or 148 will provide no output which results in the occurrence of a low signal at an output terminal 152 from the matrix 150. In effect, the signal appearing at the output 152 comprises simply an "and" combination of the outputs from each of the comparators.

In the event of an ambiguity, it may be desirable to develop other digital values or alternatively to simply record the output from the comparators 130, 141, 142, 144, 146 and 148. For example, in the event of an ambiguity, the outputs from the comparators may comprise the binary digits of a value in combination with the signal appearing at the terminal 152. Of course, a wide variety of different encoding schemes are readily useable with the system. However, the singificant consideration lies in the appreciation that the encoding operation stemming from relative magnitudes of accumulated values accomplishes a substantial reduction in the data yet preserves an identifying consideration. Although the above explanation has assumed a simply one-to-one comparative relationship, it is also apparent that other relationships may be developed for increased flexibility of the system. Specifically, for example, such relationships might query whether or not the summation of one of the values is more than twice the summation of another value. Thus, a great number of relationships are possible to formulate the basis for encoding the summation signals into a more concise digital code.

Claims

1. A system for statistical identification of a time-referenced substantially-continuous signal sequence, comprising:

timing means for defining a plurality of periods in time offset lapped relationship and within an interval of said signal sequence;

means for analyzing said signal sequence during each of said periods defined by said timing means, to provide a plurality of representative digital code signals; and

means for registering said plurality of representative digital code signals

2. A system according to claim 1 wherein said means for analyzing said signal sequence includes a digital filter means for providing digital

3. A system according to claim 1 wherein said means for analyzing includes means for comparing represented magnitudes of components of said signal

4. A system according to claim 1 further including means for comparing said plurality of representative digital code signals from said means for registering with another plurality of representative digital code signals.

5. A system according to claim 1 wherein said means for analyzing said signal sequence includes a digital filter means for providing digital values from said signal sequence and wherein said means for analyzing includes means for comparing represented magnitudes of components of said

6. A system according to claim 5 further including means for comparing said plurality of representative digital code signals from said means for registering with another plurality of representative digital code signals

7. A system according to claim 1 wherein said means for analyzing further includes means for manifesting an ambiguity in said digital code signals.