WO1999041893A1

WO1999041893A1 - Call progress tone detection

Info

Publication number: WO1999041893A1
Application number: PCT/US1999/002988
Authority: WO
Inventors: Steve Jarboe
Original assignee: Intervoice Limited Partnership
Priority date: 1998-02-12
Filing date: 1999-02-12
Publication date: 1999-08-19
Also published as: AU3289899A

Abstract

A system and a method (10) for detecting and identifying call progress tones are disclosed. Signals on a telephone line are sampled and the frequencies having the highest power are identified as signal components. The frequencies and power level for each component and the total signal power are autocorrelated to generate autocorrelation peaks for each of the three components. Signal information (13), such as autocorrelation peaks, duty cycle and pulse repetition interval is compared to a database of call progress tone templates. The call progress tone is identified by matching it to one of the database templates (14).

Description

CALL PROGRESS TONE DETECTION

TECHNICAL FIELD OF THE INVENTION

This invention relates to the detection of telephone call progress tones and more specifically to a system and method for detecting call progress tones using autocorrelation techniques on feature vectors.

BACKGROUND OF THE INVENTION

The use of call progress tones is well known in the telecommunications field. Usually, different types of tones are used to indicate line status or call progress to a caller. The tones are distinguished from each other by varying the frequency, pulse repetition interval or duty cycle of the tones or by using a sequence of multiple frequency tones. Typically, the tones indicate that a line is available to place a call (dial tone), that the telephone assigned to the number called is ringing (ring back signal), that the number called is busy (busy signal) or out of service, that the telephone circuits are busy or that the caller needs to redial using a different number sequence. In some cases, such as dial tones and busy signals, only the call progress tones are broadcast to the caller. In other cases, the tones are followed by a recorded message, such as an out of service message or directions to modify the number dialed. For example, the caller may be directed to add or delete a " 1 " or the area code to the number sequence.

The standard call progress tones are readily distinguishable to a human caller. The vast majority of people who use the telephone can generally distinguish and classify the tones even if the duty cycle, frequency or other parameters are slightly different from one call to the next or from one system to the next. However, automated telecommunications systems can have difficulty differentiating among the call progress tone variations. If an automated telecommunications system is to operate properly, it must be able to identify each call progress tone and to distinguish among the various types of tones. For example, an automated system must properly identify and distinguish a busy tone from a ring tone, otherwise the system may disconnect from the line before the call is answered.

A typical telecommunications system uses filters to detect and identify call progress tones. A group of filters are individually tuned to identify each call progress tone. Accordingly, one filter would detect a dial tone, a second filter would detect busy signals, etc. This system of filters is sufficient as long as the parameters of the tones do not vary too broadly with each call. However, if the individual types of call progress tones vary significantly, then a new filter is needed for each variation. Telephone systems around the world are no longer controlled by single entities in each country, but rather by a number of different companies in each country. In a "single presence" situation where the telephone system is run by one company, call progress tones are essentially uniform and each type of tone can be identified by the same filter on every call. However, in view of the increasing variety of proprietary telephone equipment that is in use, it can no longer be assumed that one type of call progress tone will have the same parameters on different systems in the same country. The situation is even worse when comparing tones in different countries. For example, a busy signal or ring signal may vary in frequency and duty cycle from one telephone system to another. Usually, a human caller will not have a problem identifying and distinguishing between the tones despite the variations in parameters. However, when a computer or other automated device initiates a call, the fixed parameters of the call progress tone filters may not be able to identify tones that vary from the expected parameters, or may require too much time to perform properly.

In an automated telecommunications system, a filter that is designed to detect and identify one system's busy tone might not detect a second system's busy tone if the second tone has a slightly different frequency or duty cycle. In order to detect and identify the second busy tone, prior art systems must employ a second filter that is designed for the parameters of the second signal. As a result, prior art systems require a separate call progress tone filter for every variation of each type of tone. This requires a separate filter for every system's variation of the ring tone, busy signal, out of service signal, etc. If these systems encounter a new type of tone which does not match one of the filters, then the system will not be able to identify the call progress tone and, therefore, will not be able to handle the call properly.

In addition to variations in call progress tones across different switches or telephone systems in the United States, international calling presents a similar problem. An automated calling system is likely to encounter different tones in every country called. As more countries are called, the prior art systems require more and more specially tuned filters to detect and identify the various call progress tones. This situation is costly both in the amount of time required to continuously update the system and the amount of hardware to required to add new filters to the system. For each tone variation, a filter must be set for a specific frequency. The filter must measure the time that the tone is on (pulse width) and the time that the tone is off. These parameters are then compared to the expected parameters in each filter. Slight differences in frequency or pulse width may cause a tone to be rejected by the filter that is actually intended to identify that type of tone. For example, in the United States a ring tone cycle is typically on for two seconds and off for four seconds, giving a thirty-three percent duty cycle. If a PBX or foreign telephone system generates a ring signal that is on for only three-quarters of a second and off for four seconds or a ring signal that is on for two seconds but off for only three seconds, then a filter designed to detect a standard ring signal would reject this tone because the pulse width and/or duty cycle parameters are incorrect. On the other hand, a human caller would probably have no problem identifying the tone type.

By using some type of error window the filters can be designed to accept some variation in the tones. However, if the there is too much room for error, false detections and incorrect identifications will become a problem. This could lead to call progress tones being misidentified, thereby causing the system to incorrectly handle the call. For instance, misidentifying a ring signal as a busy signal will cause the system to terminate the call before it is answered. On the other hand, misidentifying a busy signal as a ring signal will cause the system to waste time by remaining connected to the line while waiting for an answer.

Noise presents another problem for the filters. Too much noise on the telephone lines or in the system will make it difficult for the filter to distinguish the call progress tones from the background noise. In some cases, the noise or voice may randomly meet a filter's design parameters and accidently trigger the filter resulting in a misidentification of noise as a call progress tone. SUMMARY OF THE INVENTION

These and other problems and deficiencies in the prior art systems for detecting call progress tones are overcome by a system and method which uses autocorrelation techniques in place of the prior art's signal filters. In the present invention, parameters (or features) of the signal on a telephone line are sampled at periodic intervals. In each sample, the three frequency components having the highest power levels are identified. These frequency and power level components are provided to an analyzer as a feature vector. Each feature vector represents the characteristics of the signal detected on the telephone line during that sample period. A sequence of the feature vectors are used by the analyzer to autocorrelate each of the signal components. The total signal power is used to normalize the power levels over each sample. The autocorrelation generates a series of peaks corresponding to pulses on the telephone line. If there is a call progress tone on the telephone line, then the autocorrelation peaks should correspond to the pulse repetition interval of the call progress tone. The series of pulses in a call progress tone should each have approximately the same duration, frequency and power level and they should be spaced at approximately the same interval. This allows the system to distinguish and reject noise and spurious signals that have varying durations, power levels or non-uniform pulse spacing.

After a potential call progress tone is detected, then the signal parameters, such as interval and duty cycle, can be compared to a database of known call progress tone parameters. The detected call progress tone can be identified as a specific type by matching the parameters to the database. Once the call progress tone is identified, then the calling system can determine how to proceed with the call. For example, wait for an answer if a ring signal is detected, re-cue the number and try again later if a busy signal is detected or set the number aside if the call cannot be completed as dialed.

It is one feature of the present invention to provide a system which can identify individual call progress tones without requiring a specially tuned filter for all of the possible variation of each tone. It is another feature of the present invention to identify call progress tones without knowing an exact frequency, pulse repetition interval, pulse width or duty cycle for each variation of a call progress tone.

It is a further feature of the present invention to provide a system that uses a software database of known call progress tone parameters and to provide an easy method for updating the list of known call progress tone parameters. The use of an updatedable software list obviates the need for extensive use of single-function hardware filters.

It is an additional feature of the present invention to allow for the identification of both single frequency and multiple frequency call progress tones. Furthermore, multiple frequency tones can be identified either when the frequencies are transmitted simultaneously or when they are transmitted sequentially.

It is also a feature to allow for an extremely broad definition of each call progress tone and have a low probability of false detection.

The foregoing has outlined rather broadly the features and technical advantages of the present invention in order that the detailed description of the invention that follows may be better understood. Additional features and advantages of the invention will be described hereinafter. It should be appreciated by those skilled in the art that the conception and the specific embodiment disclosed may be readily utilized as a basis for modifying or designing other structures for carrying out the same purposes of the present invention. It should also be realized by those skilled in the art that such equivalent constructions do not depart from the spirit and scope of the invention as set forth in the appended claims.

BRIEF DESCRIPTION OF THE DRAWINGS

For a more complete understanding of the present invention, and the advantages thereof, reference is now made to the following descriptions taken in conjunction with the accompanying drawings, in which: FIGURE 1 is a block diagram illustrating a preferred embodiment of the present invention;

FIGURE 2 illustrates the relationship between the components of a busy signal as detected by the present invention;

FIGURE 3 illustrates the relationship between the components of warble tone as detected by the present invention;

FIGURE 4 illustrates the relationship between the components of a ring signal as detected by the present invention; and

FIGURE 5 illustrates the relationship between the components of an out of service signal as detected by the present invention.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

The preferred embodiment of the present invention is shown in FIGURE 1 as system 10. System 10 can be part of an automated telecommunications system that detects call progress tones in order to place calls. Telephone signals are provided to system 10 via line 101. Line 101 may be a direct connection to a telephone line or it may carry signals from another telecommunications system. Frequency and power estimator 11 receives the telephone signals and estimates the signal components of the telephone signal at periodic intervals. The estimation interval is selected based upon the system in which the present invention is employed. The estimation interval depends upon the processing capacity of the system and the number of samples that the system can handle. Shorter intervals will produce more samples and allow for greater accuracy. Another consideration is the frequency and duration of the expected call progress tones. For example, using a 10 millisecond interval will allow a system to distinguish between tones that vary by 10 milliseconds or more in duration. On the other hand, if a ring signal is "on" for one second and "off" for four seconds, then the 10-millisecond interval will generate 500 samples per ring cycle.

During each interval, estimator 11 will sample the frequency and power of the strongest signals on line 101. In the preferred embodiment, the three strongest signals will be sampled. However, in other embodiments, the present invention may sample more or less signals depending upon the expected call progress tone format. If the tones are expected to have no more than two frequencies, then it may be practical to only detect the frequency and power of the two strongest signals. In other cases more samples may be needed. For example, if more than three frequencies are expected during a sample or if the levels of the noise or spurious signals are near the level of the call progress signal, then more samples may be needed to facilitate accurate detection of the call progress tones and to distinguish the call progress tones from noise. Changes in the number of samples can be automatic based on certain parameters or could be timed or programmed depending upon the expected signal at any given time.

Estimator 11 can be embodied as any device that is capable of measuring the strength of the detected signals and providing data related to the signals. A series of filter banks or adaptive notch filters would serve this purpose. A device capable of performing a fast Fourier transform or signal estimation could also be used for estimator 11.

In the preferred embodiment, estimator 11 outputs the frequency and power information for the three strongest signals over lines 102a-c. Additionally, estimator 11 provides a measurement of the total power σ_x ² for the overall signal on line 101. Power σ_x ² represents the total power across the entire frequency band. The matrix of frequencies and powers for the three strongest signals along with the total power value represents a feature vector. For each sample interval a feature vector is generated representing the characteristics of that sample. In the preferred embodiment shown, the components of the feature vectors are provided to feature vector analyzer 12 via lines 102a-c and 103. In an alternate embodiment, the components may be sent over a single line, instead of multiple lines 102a-c and 103.

Analyzer 12 is a computer or some other processor that is capable of receiving and analyzing feature vector data for each sample. Analyzer 12 compares the feature vectors and detects patterns from the data. Information about the signal on line 101 is generated from these patterns. Using the feature vector data, analyzer 12 can determine the pulse repetition interval, duty cycle, average pulse power and average pulse duration. This information and any other information related to pertinent signal characteristics are provided (for example) over lines 104a-d to processor 13 which uses the information to determine which type of call progress tone, if any, has been detected on line 101.

Analyzer 12 performs an autocorrelation calculation on the sequence of feature vectors that are received from estimator 11. The autocorrelation will produce a series of autocorrelated peaks for each of the three strongest signals detected on line 101. If a call progress tone is present on line 101, then the peak autocorrelation points will correspond to the pulses in the call progress tones. For example, if the tone on line 101 is a busy signal that alternates from "on" for one-half second and "off for one-half second, the autocorrelation in analyzer 12 will produce a series of autocorrelation peaks at one second intervals. The autocorrelation peaks will correspond to the "on" pulses in the busy signal. Using the frequency of the signal and the spacing between the autocorrelation peaks, the characteristics 10

of the signal on line 101 can be compared with a list of known parameters for example from database 14 to identify the call progress tone.

By using autocorrelation techniques, the call progress tones can be distinguished from background noise. Although noise spikes may have the highest power level in some samples, the autocorrelation will reject these spurious signals since are not regularly repeating signals. In the preferred embodiment, system 10 detects the three strongest signals in each sample. Since most call progress tones do not use three frequencies, in many cases one or two of the strongest signals in each sample will be caused by noise. Any autocorrelation peaks that are detected in these noise samples will be rejected by system 10 since it is extremely unlikely that they would appear in a regularly repeating pattern. The normalized autocorrelation would be small and the power level at that frequency would be low. The noise samples will create a random series of autocorrelation peaks that will not equate to any expected call progress tone parameters.

A telecommunications system using the present invention would also maintain a table of call progress tone parameters 14. This table may be a software database so that it can be modified or upgraded easily. Parameter database 14 holds information about each type of call progress tone, such as the duty cycle of the tone, the number of frequencies expected and the duration of the pulses. Other considerations could also be loaded into the database. For example, the database could maintain information about the relationship between the frequencies in multiple frequency tones, such as whether the frequencies are "on" at the same time or at alternating times.

To detect and identify variations in the call progress tones, the database could allow for a specified degree of error in the parameters or a new model or template could be designed for each potential call progress tone variation. Using the international calling example, the primary ring signal template may be set to identify a ring tone cycle that is "on" for one second and "off' for four seconds. Additional templates could be added to the software database so that the system would also recognize one foreign country's ring signal that is "on" for only three-quarters of a second and "off' for four seconds and another country's ring signal that is "on" for one second but "off' for only three seconds. These 11

additional templates could be added by a simple database addition or with a second auxiliary database without modifying the systems hardware or adding new hardware filters.

FIGURE 2 is a series of graphs illustrating the feature vector and autocorrelation components for a typical busy signal. In this example the signal is comprised of a two frequency tone. Graphs 201a-c represent the frequencies of the three strongest signals detected on line 101 by estimator 11. Graphs 202a-c represent the corresponding power levels for the signals in lines 201a-c. The total power of the signal is shown as graph 203.

Estimator 11 samples the signal at the selected periodic interval. Initially, while the tone is off, the only signals on the line are random noise or spurious signals as shown in graphs 201c and 202c. Since this signal is caused by noise, the strongest signal detected in the initial samples will vary randomly in frequency and power during each sample interval.

When the busy signal begins and the tone first goes "on", estimator 11 detects a first signal with frequency 1 and power level 1 as shown in graphs 201a and 202a. At the same time, estimator 11 detects a second signal having frequency 2 and power level 2 as shown in graphs 201b and 202b. These signals, along with total power 203, are provided to analyzer 12. For the duration of the "on" pulse, estimator 11 will detect frequency and power 1 and 2 in each sample. When the pulse goes "off' then only the noise levels will be detected on the line. Accordingly, until the busy tone goes back "on" each sample will have a random frequency and power that are determined from the spurious noise on line 101. Graphs 204a-f represent the results of the normalized autocorrelation process in analyzer 12. Autocorrelation φ_F1 on graph 204a represents the series of autocorrelation peaks that are produced from signal 201a and φ_P1 204b is the autocorrelation of signal 202a. Analyzer 12 detects the regularly repeating series of samples that have the same frequency and power level. These samples will be shown as peaks in the autocorrelation. As illustrated, these peaks will be spaced at the same interval as the pulses on lines 201a and

202a. In a similar fashion, graph φ_F2 204c represents the autocorrelation peaks for the signal 201b and the peaks in φ_F2204c have approximately the same spacing as 201b. This is also true for autocorrelation φ_P2204d and signal 202b. 12

Autocorrelations φ_F3 204e and φ_P3 204f are the result of the random noise signals in graphs 201c and 202c. As illustrated in graph 204e and 204f, an autocorrelation of the noise signal will generate randomly spaced peaks which are not nearly as distinct as those for signals 201a,b and 202a,b. Analyzer 12 will attempt to correlate the noise samples. However, since the frequency and power vary randomly, the resulting autocorrelation peaks are unevenly spaced and lack distinct peaks.

The autocorrelation data in graphs 204a-c can be used by a telecommunication system to detect and identify the call progress tone on line 101. The pulse repetition interval for the autocorrelation peaks in graphs 204a-d can easily be determined. The intervals are consistent enough to be identified as a potential call progress tone. The duty cycle and pulse duration of the signals can also be determined using samples of the signals in graphs 201a,b and 202a,b. This information taken together is enough to accurately identify which type of call progress tone has been detected by comparing the pulse repetition interval, pulse duration and duty cycle power and frequencies to the database of call progress tones templates.

In the present example, a 50 percent duty cycle would mark the tone as a potential busy line signal. The simultaneously broadcast dual frequencies also distinguish the signal. This information alone should be sufficient to provide for a high detection rate and a low error rate for busy signals. System 10 would not need to look for a specific frequency or pulse repetition interval. As long as the significant frequencies which were detected have a duty cycle of approximately 50 percent, system 10 would detect and identify the signal components in FIGURE 2 as a representing a busy signal type of call progress tone.

If a prior art method was used to detect and identify the signals in FIGURE 2, then a separate filter for each potential frequency would be required. Other prior art methods would have a filter with a wide bandpass, which makes it highly susceptible to false detection. But in the present invention only one model of the busy signal is needed. After identifying the busy signal, the telecommunications system could disconnect the call and move the telephone number to a later position in a cue so that the number could be tried again later. 13

FIGURE 3 represents the components of a warble or alert tone on a telephone line. Instead of switching between on and off states, this tone switches between two different frequencies as shown on graphs 301a,b. Graph 301c represents the random noise and spurious signals on line 101 like graph 201c in FIGURE 2. The power levels illustrated in graphs 302a-c correspond to the signals in graphs 301 a-c. The total power of the signal is shown in graph 303. Autocorrelation of the signal components that are shown in graphs 301a-c, 302a-c and 303 produces peaks φ_F1-F3 and φ_P1._P3 as shown in graphs 304a-f

The signal components in graphs 301a,b and 302a,b have a 50% duty cycle. In the prior art systems, if these signals were at the correct frequency, they might trigger the busy signal filter even though they are clearly not a busy signal. A human caller would recognize the shifted tones of the signal. However, an automated prior art system would not be able to distinguish this call progress tone from a busy signal unless it was capable of comparing the outputs of two specific filters simultaneously to detect the alternating frequency shift.

In the present invention the signal is easily distinguishable from a busy signal. The autocorrelation peaks show that there is an obvious repeating pattern due to the call progress tone. The total power is relatively constant instead of bimodal.

FIGURE 4 illustrates a single-frequency ring signal as detected by the present invention. Graphs 401a and 402a represent the frequency and power components of the signal. Graphs 401b,c and 402b,c are the frequency and power levels for the noise and spurious signals during each sample interval and graph 403 is the total power on line 101. After autocorrelation in analyzer 12, peaks φ_F1-3 and φ_P1._P3 are generated as shown in graphs 404a-e.

As in the above examples, the present invention will clearly be able to reject the noise signal components shown in graphs 404b, c since they do not correspond to any fixed pulse repetition interval. However, in the case of a ring signal the telecommunications system may have a slightly more difficult time identifying the tone due to the longer ring signal cycle. Typically, a ring signal has a duty cycle on the order of 20 percent and each cycle may be 4 or 5 seconds long. As a result, the system will require more time to detect and identify a ring signal compared to other call progress tones. 14

To identify the ring tone faster, the system can use other tone parameters in addition to autocorrelated peaks 404a. For instance, each ring signal pulse should have the same duration. After two cycles system 10 should be able to compare pulse widths to help identify the autocorrelated peaks as a call progress tone. If the pulses have the same or nearly identical duration and power levels, then there is a higher probability that they are part of a call progress tone. Once the pulses are determined to be a call progress tone, then the duty cycle and the duration of the on and off periods can be used to classify the tone as a ring signal.

FIGURE 5 is an example of the signal components of a special call progress tone, such as a "number out of service" or "call cannot be completed as dialed" advisories.

Usually, this type of message is preceded by a multiple frequency tone. Although this type of tone might be broadcast only once, it is possible to identify it using the present invention. Graphs 501 a-c and 502a-c illustrate a series of multiple frequency tones. The total power is illustrated in graph 503. Following autocorrelation in analyzer 12, the autocorrelation peaks are shown in graphs 504a-c.

The signal components represented in graphs 504a-c clearly have no pulse repetition or duty cycle and, therefore, can be distinguished from the busy, ring or alert types of tones. However, by comparing the signal components a pattern can be detected and the tone can be identified. The duration of each frequency and the signal strength of each frequency in graphs 501 a-c and 502a-c will be approximately equal. However, the frequency of each tone is different and each tone is at a sequentially higher frequency. Using the overall characteristics of each component the special call progress tone can be identified and handled properly by system 10.

By identifying this special type of call progress tone, system 10 would know to skip this number and move to another number. Otherwise, since it is not a busy or alert type of tone, the telecommunications system might remain connected to the line until a system timeout prompted it to move to another number. The system could also be programmed to attempt to alter the number dialed in response to a special call progress tone. For example, the system could add or delete a "1" or an area code to the number and dial the number again. If these preset modifications did not result in a connection, then the number could be 15

marked and set aside so that a human operator could listen to the message and handle the number accordingly.

The present invention could be adapted to detect and identify more complex call progress tones in the same manner that the special tone of FIGURE 5 is identified. Tones having more frequencies or more complicated arrangements of the pulses could also be identified. The database of potential call progress tones would only have to be updated for the system to operate properly.

Although the present invention and its advantages have been described in detail, it should be understood that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims

16WHAT IS CLAIMED IS:

1. A system for detecting and identifying signals in a communication network comprising: means for receiving signals in said communication network, wherein said signals are comprised of one or more components; means for sampling said received signals at periodic intervals; means for measuring one or more parameters of said components in said sampled signals; means for autocorrelating a series of said one or more measured parameters to detect a pattern in said signal; and means for comparing said detected pattern with known signal characteristics to identify a specific type of signal.

2. The system of claim 1 further comprising: means for maintaining a list of known signal characteristics.

3. The system of claim 2 further comprising: means for changing said list of said known characteristics.

4. The system of claim 3 wherein said changing means further comprises: means for updating said known characteristics list using detected characteristics.

5. The system of claim 1 wherein said components comprise one or more signals having different frequencies.

6. The system of claim 1 wherein said parameters comprise a frequency and a power level. 17

7. The system of claim 6 wherein a first component comprises a first frequency and a first power level and wherein a second component comprises a second frequency and a second power level.

8. The system of claim 7 wherein at least one of said components is a pulse.

9. The system of claim 8 wherein said pulse comprises an "on" time.

10. The system of claim 8 wherein said pulse comprises an "off time.

11. The system of claim 8 wherein said parameters of said pulse component further comprise a duty cycle.

12. The system of claim 8 wherein two or more of said components are pulses that are transmitted at the same time.

13. The system of claim 8 wherein two or more of said components are pulses that are transmitted sequentially.

14. The system of claim 1 wherein said communication network is a telephone system and said signals are received over a telephone line.

15. The system of claim 14 wherein said signals received over said telephone line are call progress tones.

16. The system of claim 1 wherein said component parameters compared in said comparing means comprise a pulse repetition interval.

17. The system of claim 1 wherein said component parameters compared in said comparing means comprise a duty cycle. 18

18. The system of claim 1 wherein said component parameters compared in said comparing means comprise a pulse width.

19. The system of claim 1 wherein said means for maintaining a list of known signal characteristics comprises a software database.

20. The system of claim 19 wherein said list of known signal characteristics comprises a list of call progress tone characteristics.

21. A method for detecting and identifying signals in a communication network comprising the steps of: receiving signals from said communication network, wherein said signals are comprised of one or more components; sampling said received signals at periodic intervals; measuring one or more parameters of at least one of said signal components in said sampled signals; correlating a series of said one or more component parameters to detect a pattern in said signal; and comparing the component parameters corresponding to said detected pattern with said known signal characteristics to identify a specific type of said signal.

22. The method of claim 21 further comprising the step of: maintaining a list of known signal characteristics.

23. The method of claim 22 further comprising the step of: changing said list of said known signal characteristics.

24. The method of claim 23 wherein said changing step comprises updating said list of said known signal characteristics depending upon detected characteristics. 19

25. The method of claim 22 wherein said parameters of said components comprise a frequency and a corresponding power level.

26. The method of claim 25 wherein a first component comprises a first frequency and a first power level and wherein a second component comprises a second frequency and a second power level.

27. The method of claim 26 wherein at least one of said components is a pulse.

28. The method of claim 27 wherein said pulse comprises an "on" time.

29. The method of claim 27 wherein said pulse comprises an "off time.

30. The method of claim 27 wherein said parameters of said pulse component further comprise a duty cycle.

31. The method of claim 30 wherein two or more of said components are pulses that are transmitted at the same time.

32. The method of claim 31 wherein two or more of said components are pulses that are transmitted sequentially.

33. The method of claim 21 wherein said communication network is a telephone system and said signals are received over a telephone line.

34. The method of claim 33 wherein said signals are call progress tones in said telephone system.

35. The method of claim 21 wherein said component parameters compared in said comparing step comprise a pulse repetition interval. 20

36. The method of claim 21 wherein said component parameters compared in said comparing means comprise a duty cycle.

37. The method of claim 21 wherein said component parameters compared in said comparing means comprise a pulse width.

38. The method of claim 21 wherein said list of known signal characteristics is maintained in a software database.

39. The system of claim 21 wherein said list of known signal characteristics comprises a list of call progress tone characteristics.

40. A method of processing pulses in a communication system comprising the steps of: receiving signals from said communication network, wherein said signals are comprised of one or more pulses; sampling said signals at periodic intervals; measuring at least one parameter of said sampled signals; and correlating a series of said sampled signal parameters to detect said pulses.

41. The method of claim 40 further comprising the steps of: measuring characteristics of said detected pulses; and comparing said measured characteristics of said pulses with a list of known pulse characteristics to identify a specific type of said pulses.

42. The method of claim 41 wherein said communication system is a telephone network and said signals are received from a telephone line.

43. The method of claim 42 wherein said pulses correspond to call progress tones. 21

44. The method of claim 41 wherein said pulses comprise two or more simultaneous pulses each having a different frequency.

45. The method of claim 41 wherein said pulses comprise a series of two or more pulses each having a different frequency.

46. The method of claim 41 wherein said list of known pulse characteristics is maintained in a software database.

47. A system for detecting pulses in a communication system comprising: means for receiving signals, wherein said signals comprise at least one of said pulses; means for sampling said received signals at periodic intervals; means for measuring one or more characteristics of at least one component of said sampled signals; and correlating a series of said characteristic measurements to detect said pulses.

48. The system of claim 47 further comprising: means for measuring parameters of said detected pulses; means for identifying a specific type of said pulses using said measured parameters of said detected pulses.

49. The system of claim 48 wherein each of said at least one components corresponds to one or more of said pulses.

50. The system of claim 49 wherein said characteristics of said at least one components comprise a frequency and a power level of said one or more pulses.

51. The system of claim 50 wherein said parameters of said detected pulses comprise a duty cycle.