Real-time monitoring of perceived quality of packet voice transmission

1. A method, comprising:

obtaining, by a network device, a reference matrix based on estimates of perceived audio quality of at least portions of one or more first packetized audio messages, the reference matrix modeling values of a plurality of characteristics associated with a particular quality level;

receiving, by the network device, one or more second packetized audio messages;

evaluating, by the network device, at least portions of one or more of the one or more second packetized audio messages to obtain measurements associated with the plurality of characteristics;

creating, by the network device, a test matrix using the obtained measurements; and

comparing, by the network device, the test matrix and the reference matrix to predict a quality level associated with the one or more second packetized audio messages.

2. The method of claim 1 , wherein the network device is in a first portion of a communications network, the first portion comprising a public switched telephone network (PSTN), and receiving the one or more second packetized audio messages comprises receiving the one or more second packetized audio messages from a second portion of the communications network, the second portion comprising a packet switched network.

3. The method of claim 1 , wherein comparing the reference matrix and the test matrix comprises:

creating a reference pattern matrix, the reference pattern matrix corresponding to differences between the test matrix and the reference matrix, and

comparing the reference pattern matrix to a threshold quality level.

4. The method of claim 1 , wherein comparing the reference matrix and the test matrix comprises:

creating a cumulative matrix, including summing each row of the test matrix, where each element in a row in the cumulative matrix is a sum of all preceding elements in a corresponding row of the test matrix;

deriving a cumulative distribution function (CDF) matrix based on the cumulative matrix, including assigning, to each of element of the CDF matrix, a value of a corresponding one of the elements in the cumulative matrix divided by a largest value in a corresponding one of the rows in the cumulative matrix; and

comparing each of the elements of the CDF matrix with a corresponding element of the reference matrix to create a reference pattern matrix,

wherein one of the elements of the reference pattern matrix is populated with a zero when either:

the corresponding element of the CDF matrix is populated with zero, or

the corresponding element of the CDF matrix is populated with a value greater than a prespecified value, and

wherein one of the elements of the reference pattern matrix is populated with a 1 when the corresponding element of the CDF matrix is populated with a value greater than zero and less than or equal to the prespecified value.

5. The method of claim 1 , wherein evaluating the at least portions of one or more of the one or more second packetized audio messages further comprises evaluating a characteristic of a reconstituted digital representation of the at least portions of one or more of the one or more second packetized audio messages.

6. The method of claim 1 , where the plurality of characteristics comprise at least one of:

a dropped packet rate (DPR), or

a round trip packet latency (RTL).

7. The method of claim 1 further comprising:

evaluating, by a network device, the at least portions of the one or more first packetized audio messages to obtain sample measurements of the plurality of characteristics;

processing, by the network device, the sample measurements to produce the estimates of perceived audio quality; and

creating, by the network device, the reference matrix based on the estimates of perceived audio quality of the at least portions of one or more first packetized audio messages.

8. The method of claim 1 further comprising:

comparing, by the network device, the quality level and a threshold quality level; and

generating, by the network device, an error message in response to the quality level being less than the threshold quality level.

9. The method of claim 8 , wherein the one or more second packetized audio messages are routed on a first path over a communications network, and further comprising causing, based on generating the error message, one or more third packetized audio messages to be routed on a second path over the communications network, the second path being different from the first path.

10. The method of claim 8 , wherein generating the error message occurs in near real time of receiving the one or more second packetized audio messages.

11. The method of claim 8 , wherein generating the error message comprises:

generating a description associated with comparing the quality level to the threshold quality level; and

generating a cumulative distribution function (CDF) matrix, where the CDF matrix is derived from the test matrix and includes at least one non-zero entry indicating that the quality level is below the threshold quality level.

12. A method comprising:

receiving, by a network device, a packetized audio message;

evaluating, by the network device, at least portions of the packetized audio message to obtain measurements associated with a plurality of characteristics, each of the plurality of characteristics affecting user perception of audio quality, the evaluating of the at least portions of the packetized audio message comprising:

evaluating a characteristic of a reconstituted digital representation of the packetized audio communication, including:

identifying a difference between successive samples in the reconstituted digital representation, and

calculating at least one of:

a raw distortion measurement associated with the reconstituted digital representation,

a normalized score corresponding to the raw distortion measurement, or

a kurtosis value of a distribution of the difference;

predicting, by the network device and based on the measurements, a quality level associated with the packetized audio message.

13. The method of claim 12 further comprising:

comparing, by the network device, the quality level and a threshold quality level; and

generating, by the network device, an error message in response to the quality level being less than the threshold quality level.

14. An apparatus comprising:

a computing device configured to:

obtain, based on estimates of perceived audio quality of at least portions of one or more first packetized audio messages, a reference matrix that models values of the plurality of characteristics associated with a particular quality level;

create a test matrix using measurements of at least portions of one or more second packetized audio messages, the measurements being associated with the plurality of characteristics;

compare the test matrix and the reference matrix to produce a comparison result; and

predict, based on the comparison result, a quality level associated with the packetized audio message.

15. The apparatus of claim 14 , where, the plurality of characteristics include at least one of a dropped packet rate (DPR) or a round trip packet latency (RTL).

16. The apparatus of claim 14 , wherein the computing device is further configured to generate an error message when the quality level is below a threshold level.

17. The apparatus of claim 16 , wherein the error message includes a cumulative distribution function (CDF) matrix, and the CDF matrix is derived from the test matrix and includes at least one non-zero entry that indicates that the quality level is predicted as below the threshold quality level.

18. The apparatus of claim 14 , where the computing device, when predicting the quality level, is further configured to:

create a cumulative matrix, including summing each row of the test matrix, where each element in a row in the cumulative matrix is a sum of all preceding elements in a corresponding one of the rows of the test matrix;

derive a cumulative distribution function (CDF) matrix based on the cumulative matrix, including assigning, to each of the elements of the CDF matrix, a value of a corresponding one of the elements in the cumulative matrix divided by a largest value in a corresponding one of the rows in the cumulative matrix; and

compare each of the elements of the CDF matrix with a corresponding element of the reference matrix to create a reference pattern matrix,

wherein one of the elements of the reference pattern matrix is populated with zero when either

the corresponding element of the CDF matrix is populated with zero, or

the corresponding element of the CDF matrix is populated with a value greater than a prespecified value, and

wherein one of the elements of the reference pattern matrix is populated with 1 when the corresponding element of the CDF matrix is populated with a value between zero and the prespecified value.

19. The apparatus of claim 14 , where the computing device, when creating the test matrix, is further configured to evaluate a characteristic of a reconstituted digital representation of the at least portions of the one or more second packetized audio messages.

20. The apparatus of claim 14 , where the processor is further configured to generate an error message in response to determining that the quality level is below a threshold level.

21. The apparatus of claim 14 wherein the processor is further configured to:

evaluate the at least portions of the one or more first packetized audio messages to obtain sample measurements of the plurality of characteristics;

process the sample measurements to produce the estimates of perceived audio quality; and

create, based on the estimates of perceived audio quality of the at least portions of one or more first packetized audio messages, the reference matrix.

22. An apparatus comprising:

a computing device configured to:

create a test matrix using measurements of a packetized audio message associated with a plurality of characteristics by:

evaluating a characteristic of a reconstituted digital representation of the packetized audio communication,

identifying a difference between successive samples in the reconstituted digital representation, and

calculating at least one of:

a raw distortion measurement associated with the reconstituted digital representation,

a normalized score corresponding to the raw distortion measurement, or

a kurtosis value of a distribution of the difference; and

predict a quality level associated with the packetized audio message, where predicting the quality level includes comparing the test matrix to a reference matrix that models values of the plurality of characteristics associated with a particular quality level.

23. A non-transitory computer-readable medium having instructions stored thereon configured to cause a computing device to perform operations, the operations comprising:

obtaining a reference matrix based on estimates of perceived audio quality of at least portions of one or more first packetized audio messages, the reference matrix modeling values of a plurality of characteristics associated with a particular quality level;

creating a test matrix using measurements of at least portions of one or more second packetized audio messages associated with the plurality of characteristics;

predicting a quality level associated with the at least portions of one or more second packetized audio messages by comparing the test matrix to the reference matrix.

24. The non-transitory computer-readable medium of claim 23 , wherein the plurality of characteristics comprise at least one of:

a dropped packet rate (DPR); or

a round trip packet latency (RTL).

25. The non-transitory computer-readable medium of claim 23 , wherein the plurality of characteristics comprise a characteristic of a reconstituted digital representation of the one or more second packetized audio messages.

26. The non-transitory computer-readable medium of claim 23 , wherein the one or more second packetized audio messages are routed on a first path over a network, and the operations further comprise routing, in response to generating the error message, one or more third packetized audio messages on a second path over the network, the second path being different from the first path.

27. The non-transitory computer readable medium of claim 23 , wherein the operations further comprise:

evaluating the at least portions of the one or more first packetized audio messages to obtain sample measurements of the plurality of characteristics;

processing the sample measurements to produce the estimates of perceived audio quality; and

creating the reference matrix based on the estimates of perceived audio quality of at least portions of the one or more first packetized audio messages.

28. The non-transitory computer-readable medium of claim 27 , wherein the operations to generate the error message further comprise:

generating a description associated with comparing the quality level to the threshold quality level; and

generating a cumulative distribution function (CDF) matrix, wherein the CDF matrix is derived from the test matrix and includes at least one non-zero entry indicating that the quality level is below the threshold quality level.

29. A non-transitory computer-readable medium having instructions stored thereon configured to cause a computing device to perform operations, the operations comprising:

create a test matrix using measurements of at least portions of one or more packetized audio messages associated with a plurality of characteristics, the plurality of characteristics comprising a characteristic of a reconstituted digital representation of the packetized audio communication that comprises:

a difference between successive samples in the reconstituted digital representation, and

at least one of:

a raw distortion measurement associated with the reconstituted digital representation,

a normalized score corresponding to the raw distortion measurement, or

a kurtosis value of a distribution of the difference;

predict a quality level associated with the packetized audio message by comparing the test matrix to a reference matrix that models values of the plurality of characteristics associated with a particular quality level.

30. The non-transitory computer readable medium of claim 23 , wherein the operations further comprise generating an error message in response to determining that the quality level is below a threshold quality level.