Data processing: speech signal processing – linguistics – language – Speech signal processing – Application
Reexamination Certificate
1998-06-04
2001-04-10
Dorvil, Richemond (Department: 2741)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Application
C704S201000
Reexamination Certificate
active
06216105
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to the processing of signals used in telecommunications and more particularly to the transmission by digital means of customer generated signals.
BACKGROUND OF THE INVENTION
In a digital transmission system, the signals from the customer are customarily grouped into 8 binary digits, called 8-bit words. As a matter of convention, each bit of the 8-bit word is numbered. The most significant bit, the first bit to arrive, is numbered Bit-
0
, followed by Bit-
1
, Bit-
2
, and so on. The least significant bit is Bit-
7
. For transmission over the telecommunications network, these 8-bit words are then combined in a serial form with 8-bit words from other customers. Thus each customer's signal occupy a period of time within a transmission line, which is shared with other customers. The time for each customer is called a “time-slot”. This method of combining one customer's signal with others is called “time division multiplexing”. For digital transmission systems used in North America, 24 such 8-bit words are combined into a transmission system called T1. Customers usually send a series of 8-bit words, usually 8000 times a second. These words are then serially combined with 8-bit words from other customers also at 8000 times a second. The series of 8-bit words from one customer are called a channel, or a digital communications channel.
When the T1 transmission system, also called a T1 line, reaches a juncture, called a node, the 24 channels of 8-bit words may be required to be processed for the purpose of switching. For example, a customer's signal arriving on a T1 line occupying Channel 3, may be sent on its way on a second T1 line occupying Channel 23. This process is called digital switching.
If the customer's signal is data, such as an inventory report, and is in the form of a series of 8-bit words, then the process of digital switching is relatively simple in concept. When the customer's 8-bit word arrives on a first T1 line, according to the arriving channel assignment, it is stored in a temporary memory. When it is time for the customer's 8-bit word to be sent on its way on a second T1 line, the stored word is inserted at the proper time slot according to the channel assignment. However, if the customer's signal is not data but rather a voice conversation transformed into digital signals, called digitized voice, then there are other aspects of the digital switching which must be considered. This is the signaling information.
Signaling relates to the supervision aspects of a voice conversation. For example, when a person lifts up the phone to make a telephone call, the information that the customer has lifted the phone, i.e., off-hook, must be transmitted to the far end. To the person receiving the call, the ring of the telephone is another example of signaling information. For the T1 transmission system, signaling is sent along with the digitized voice in combined 8-bit words. This process is called “robbed bit signaling”. In 5 of every 6 8-bit words sent, the 8-bit words are faithfully transmitted and reproduced at the far end. However, in 1 of every 6 words sent, the least significant bit (Bit-
7
) of the 8-bit word is substituted with the signaling bit (i.e., A bit). The result is that the customer's voice is digitized into 8-bit words 5 out of 6 times, and digitized into 7-bit words 1 out of 6 times. In digitized voice, more bits represents greater fidelity in the overall system. In the T1 system, this method results in a fidelity commonly known as 7 ⅚ bit fidelity.
When the digitized voice is to be switched, the signaling information must be extracted from the incoming T1 line and reinserted into the outgoing T1 line. This is because the location of the 1 out of 6 words, where the signaling bit is inserted, is not determined by each channel, but by the timing of the T1 transmission line. In many cases, the reinserted signaling bit on the outgoing T1 line is at a different word when compared to the incoming bit stream. Here is where procession for data and for digitized voice must be different. For data, no signaling processing should take place. Otherwise, customer's data would be mutilated due to moving what is thought to be the signaling bit into a different place. For voice, a signaling process should take place to preserve the transmission of the signaling information. Thus it is important for the system administrator to know which channel is carrying data, and which channel is carrying voice. This knowledge is obtained usually by searching the records for the original customer order whether data or voice is requested. Another way is to examine the actual 8-bit words being transmitted and make a determination, based on the characteristic difference between data and digitized voice. Such prior methods are time-consuming and unreliable.
Thus, it is desirable to provide an approach for automatically differentiating between data and voice digital signals to overcome the drawbacks of the prior art.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide an automatic device for determining whether the customer's digital signals, in the form of a stream of 8-bit words, represent data, or represent digitized voice. Essential to the present invention is the assumption that if the signal is indeed digitized voice, signaling information, in the format of “robbed-bit-signaling” would also be present. Otherwise, for data, no signaling information will be found.
The basis of this automatic determination is the observation that (a) both data and digitized voice tend to change rapidly, with each consecutive 8-bit word different from the previous, and (b) signaling tends to change slowly, following the motion of the users hand, or the ringing cadence. More precisely, both data and digitized voice may change 8000 times a second, or every 0.125 milliseconds, while signaling information changes at most 20 times a second, or every 50 milliseconds. Thus it is possible, by examining the rate of change of the supposed signaling bit i.e., the bit located at bit
7
, to determine whether it is indeed a signaling bit i.e., whether bit-
7
is in fact the A bit. If it is determined to be a signaling bit, then by extension, the 8-bit words would represent digitized voice. If such a slowly changing signaling bit is not found, then by extension, the 8-bit words would represent data.
In order to simplify this embodiment of the present invention, when consecutive words need to be compared as to whether they change, only bits in position
6
(Bit-
6
) in each consecutive 8-bit word are examined to determine whether it changes.
In order to improve the accuracy of the determination of the signaling bit, the present invention proposes that each such test of the signaling bit be done twice and at different times. Only when the result of both tests agree would the determination be accepted.
In a further refinement, further cases are considered where both data and digitized voice may be idle. This is when no data is being sent, or when the digitized voice signal represents silence. In both cases, each consecutive 8-bit word may be the same idle word, and thus neither the supposed A bit, nor the words would be changing. When this condition is detected, no determination would be made.
In summary, when the supposed signaling bit A is found to be changing rapidly, then the signal is determined to be DATA. When the supposed signaling bit A is found to be not changing while the rest of the bits are changing rapidly, then the signal is determined to be VOICE. If neither of the above conditions is detected, then no conclusion is reached until the next test.
This test, of whether the signal is data or digitized voice, is repeated periodically, say once every minute. Thus, when the customer abruptly changes the signal from data to voice or vice versa, the system may not know of the change until some time later. Still, this automatic device of determination is faster than manual methods.
Bacon & Thomas PLLC
Dorvil Richemond
Loop Telecommunication International, Inc.
LandOfFree
Method and device for differentiating between data and voice... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and device for differentiating between data and voice..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and device for differentiating between data and voice... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2446701